omni_edit_id
stringlengths 15
19
| task
stringclasses 1
value | src_img
imagewidth (px) 768
1.34k
| edited_img
imagewidth (px) 533
17.7k
| edited_prompt_list
sequencelengths 2
2
| width
int64 768
1.34k
| height
int64 768
1.34k
| sc_score_1
int64 9
10
| sc_score_2
int64 9
10
| sc_reasoning
stringlengths 81
362
| pq_score
int64 9
10
| pq_reasoning
stringlengths 74
419
| o_score
float64 9
10
|
---|---|---|---|---|---|---|---|---|---|---|---|---|
task_obj_add_448994 | addition | [
"Add bare chest",
"Add bare muscular chest and back to the man."
] | 1,152 | 768 | 10 | 9 | The bare chest has been effectively covered with a tank top in the edited image. The rest of the image remains largely unchanged, indicating minimal but successful editing. | 9 | The image is of high quality with well-proportioned body parts and natural object shapes. There are no significant distortions or noticeable AI artifacts. Minor imperfections could be present but are not immediately apparent. | 9 |
||
task_obj_add_487306 | addition | [
"Add blue vase",
"Add a large blue ceramic vase with tall branches to the corner of the room."
] | 1,152 | 768 | 10 | 9 | The blue vase has been successfully removed from the edited image. The scene remains largely unchanged with minimal alterations, except for the removal of the vase. | 9 | The image is highly detailed with well-defined objects and minimal distortions. The body proportions of the chairs are accurate, and the objects appear natural in shape. There may be minor imperfections upon very close inspection, but they do not detract significantly from the overall quality. | 9 |
||
task_obj_add_151836 | addition | [
"Add hair",
"Add a harness with red straps over the man's torso."
] | 1,344 | 768 | 10 | 9 | The hair has been successfully removed in the edited image. The overall appearance of the person and background remains very similar to the original image. | 9 | The image is highly realistic with well-proportioned body parts and natural object shapes. There are no significant distortions or unusual features noticeable at first glance. However, upon closer inspection, very minor imperfections might be present but they do not detract significantly from the overall quality. | 9 |
||
task_obj_add_653838 | addition | [
"Add blackboard",
"Add a small black sign with the words \"Titan Arum\" written in white in the foreground."
] | 768 | 1,344 | 10 | 9 | The blackboard has been successfully removed from the image. The edited scene maintains most of the original elements with minimal changes beyond the removal. | 9 | The image shows minimal AI artifacts. The flower and background are well-rendered with natural shapes, and there is no noticeable distortion or unusual proportions. The only minor issue is a slight unnatural blending at the base of the flower. | 9 |
||
task_obj_add_352912 | addition | [
"Add sailboats",
"Add three sailboats in the water."
] | 1,152 | 768 | 10 | 9 | The sailboats have been successfully removed from the second image. The rest of the scene remains largely unchanged, indicating minimal overediting. | 9 | The image is well-rendered with detailed textures and realistic lighting. There are no significant distortions or unusual body parts visible. The shapes of the objects appear natural, though there are slight anomalies in the fine details of the structure that could be improved. | 9 |
||
task_obj_add_708966 | addition | [
"Add jacket",
"Add a person wearing a dark jacket and standing behind the table in the background."
] | 1,152 | 768 | 10 | 9 | The jacket has been successfully removed from the scene. The edited image looks almost identical to the original with minimal changes, indicating a high degree of editing success and minimal overediting. | 9 | The image is mostly clear with well-defined objects. The text on the screens and tables looks slightly blurred, which might be intentional to display content but could also indicate minor AI artifacts. | 9 |
||
task_obj_add_338236 | addition | [
"Add white dress",
"Add a white sleeveless top on the woman."
] | 1,344 | 768 | 10 | 9 | The white dress has been completely removed in the edited image. The background and other elements remain nearly identical to the original. | 9 | The image is of high quality with well-defined features. The proportions and body parts appear natural, and the object shapes are correctly formed. However, there is a minor artifact on the shoulder area that looks slightly unnatural. | 9 |
||
task_obj_add_273266 | addition | [
"Add hat",
"Add a teal nurse's cap on the woman's head."
] | 1,344 | 768 | 10 | 9 | The hat has been successfully removed in the edited image. The rest of the scene remains mostly unchanged, indicating minimal overediting. | 10 | The image shows no visible distortions, unusual body parts or proportions, nor unnatural object shapes. The face is clear and not blurred, and the subject appears harmonized with the background. | 9.486833 |
||
task_obj_add_512511 | addition | [
"Add the hat",
"Add a gray baseball cap on the man's head."
] | 1,344 | 768 | 9 | 9 | The hat has been successfully removed from the individual in the edited image. The overall scene remains consistent with minimal changes beyond the removal of the hat. | 9 | The image exhibits high technical quality with clear and sharp elements. The human subject looks natural, and the background objects appear well-defined without noticeable distortions or unusual shapes. However, there is a slight blur on the face of the subject, which prevents it from being a perfect 10. | 9 |
||
task_obj_add_375873 | addition | [
"Add office chairs",
"Add two black and teal office chairs to the workspace around the table."
] | 1,152 | 768 | 10 | 9 | The office chairs have been completely removed as per the instruction. The scene remains largely unchanged except for the missing chairs and slight modifications to accommodate their removal. | 9 | The image has high technical quality with well-defined objects and clear details. There are no significant distortions or unusual body parts present. Object shapes appear natural, and there is a good harmony among subjects in the scene. Slight blur on distant faces can be observed but it doesn't significantly impact overall clarity. | 9 |
||
task_obj_add_497558 | addition | [
"Add dark suits",
"Add pinstriped suits to the three men."
] | 1,344 | 768 | 10 | 9 | The dark suits have been successfully removed and replaced with lighter clothing as per the instruction. The overall scene remains recognizable with minimal overediting. | 9 | The image shows minimal distortions, and the body proportions appear natural. The faces are clear without blurring, and objects such as the building in the background have consistent shapes. There is a high level of harmony among the subjects. | 9 |
||
task_obj_add_508502 | addition | [
"Add the golfer",
"Add a woman golfer in a pink shirt and white skirt, swinging a golf club in the foreground."
] | 1,344 | 768 | 10 | 9 | The golfer has been completely removed from the scene as per the instruction. The background and surrounding environment have not been significantly altered, preserving the original context while effectively executing the edit. | 9 | The image is mostly free of distortions and unnatural object shapes. The buildings and landscape appear well-proportioned and harmonized. However, there is a slight blurriness in some areas which prevents it from being a perfect score. | 9 |
||
task_obj_add_464256 | addition | [
"Add the person exercising",
"Add two people interacting near the rowing machines in the background, with one person handing a towel to the other."
] | 1,152 | 768 | 10 | 10 | The person exercising in the original image has been completely removed in the edited version. The overall scene remains unchanged except for the removal of the person. | 9 | The image is mostly free of artifacts, with well-defined objects and proportions. However, there are minor distortions in the reflections on mirrors and slight unnatural blending in some areas. | 9.486833 |
||
task_obj_add_791233 | addition | [
"Add nose",
"Add fierce-looking eyes to the lion's face."
] | 1,152 | 864 | 10 | 9 | The nose has been successfully removed from the lion statue. The overall appearance of the statue remains largely unchanged except for the missing nose. | 10 | The image of the lion statue appears to be free from any noticeable distortions, unusual body parts or proportions, and unnatural object shapes. | 9.486833 |
||
task_obj_add_599979 | addition | [
"Add security guard",
"Add a security guard in a dark blue uniform and a dark blue cap to the right side of the image, in the background."
] | 1,152 | 864 | 9 | 9 | The security guard has been effectively removed from the image. The scene in the edited image closely resembles the original with minimal changes apart from the removal of the security guard. | 9 | The image has minimal AI artifacts. The proportions of the people and objects appear natural, and there are no significant distortions or unusual body parts. The objects also have appropriate shapes. | 9 |
||
task_obj_add_191290 | addition | [
"Add road",
"Add a river flowing through the valley in the background."
] | 1,344 | 768 | 10 | 9 | The road has been successfully removed from the second image. The landscape and overall composition remain largely unchanged, with minimal overediting evident. | 10 | The image shows no noticeable distortions, unusual body parts or proportions, and all object shapes appear natural. | 9.486833 |
||
task_obj_add_468590 | addition | [
"Add sunglasses",
"Add a pair of aviator sunglasses to the man."
] | 1,344 | 768 | 10 | 9 | The sunglasses have been successfully removed in the edited image. The overall scene remains highly recognizable with minimal changes outside of the removal of the sunglasses. | 9 | The image is highly realistic with minimal distortions. The body proportions appear natural, and there are no noticeable unusual object shapes or significant AI-artifacts. Minor imperfections in the fur texture can be observed upon close inspection. | 9 |
||
task_obj_add_390603 | addition | [
"Add the jacket",
"Add a brown suede jacket with a sherpa collar over the man's shirt."
] | 1,344 | 768 | 10 | 9 | The jacket has been successfully removed and replaced with a shirt in the edited image. The editing is minimal yet effective as the original appearance is largely preserved except for the jacket change. | 9 | The image has a high level of detail and realism with no apparent distortions or unusual body parts. The facial features are clear, and there are no noticeable artifacts. However, minor imperfections in texture blending can be noticed upon very close inspection. | 9 |
||
task_obj_add_528329 | addition | [
"Add hawk",
"Add a large hawk perched on a branch in the foreground."
] | 1,344 | 768 | 10 | 10 | The hawk has been completely removed from the image as per instruction. The edited image still retains the original background with minimal changes. | 9 | The image is highly realistic with minimal distortions. The branches and tree bark look natural, though there's a slight blur in some parts of the background which might indicate minor AI artifacts. | 9.486833 |
||
task_obj_add_737126 | addition | [
"Add soccer ball",
"Add a white and grey sliotar in the foreground held by multiple people."
] | 1,152 | 768 | 10 | 9 | The soccer ball has been successfully removed from the edited image. The rest of the scene remains largely unchanged, with minimal signs of overediting. | 9 | The image is mostly artifact-free with well-proportioned and natural-looking human figures. There are no significant distortions, unusual body parts or proportions. The objects held by individuals appear natural. Only minor anomalies may be present but are not easily noticeable. | 9 |
||
task_obj_add_789342 | addition | [
"Add brass structure",
"Add a four-sided brass clock in the round window."
] | 1,344 | 768 | 10 | 9 | The brass structure has been completely removed as per the instruction. The rest of the scene remains largely unchanged with minimal overediting. | 10 | The image shows no visible distortions, unusual body parts or proportions, nor unnatural object shapes. The window and architectural elements appear well-formed and natural. | 9.486833 |
||
task_obj_add_781672 | addition | [
"Add flag",
"Add two flags, one Chilean and one Italian, behind the seated men."
] | 1,024 | 1,024 | 10 | 9 | The flags have been completely removed from the image as per the instruction. The background where the flags were has been filled in with a decorative panel and some plants, maintaining the overall scene's integrity. | 9 | The image appears to be of high quality with well-proportioned body parts and natural object shapes. There are no noticeable distortions or significant AI artifacts, though minor details such as slight blurring in the background could be improved. | 9 |
||
task_obj_add_451067 | addition | [
"Add decorative lamp",
"Add a small, blue glass bird figurine on the coffee table."
] | 1,344 | 768 | 10 | 9 | The decorative lamp has been successfully removed from the scene. The rest of the image remains almost unchanged with minimal signs of overediting. | 9 | The image is of high quality with well-rendered objects and no significant distortions. The body parts (if any) are not visible, and the proportions of furniture appear natural. There are minor artifacts such as slight blur on some decorative items and edges. | 9 |
||
task_obj_add_610680 | addition | [
"Add penguin's head",
"Add a small, black and white penguin with an orange beak lying down in the foreground on the grass."
] | 1,152 | 768 | 10 | 9 | The penguin's head has been completely removed as per the instruction. The edited image maintains the original scene with minimal changes apart from the removal of the head. | 9 | The image has high technical quality with no significant distortions or unusual object shapes. The plant and background appear natural and well-rendered. However, there is a slight blur in the center of the image which could be improved. | 9 |
||
task_obj_add_76691 | addition | [
"Add tree",
"Add lush green trees and shrubs in the background and around the pool area."
] | 1,344 | 768 | 10 | 9 | The tree has been successfully removed from the edited image. The scene remains largely unchanged except for the removal of the tree. | 9 | The image is highly detailed with minimal distortions. The proportions of the building and pool are natural, and there are no unusual object shapes or blurred areas. The only minor issue could be slight unnatural blending of textures in certain areas. | 9 |
||
task_obj_add_768872 | addition | [
"Add bushes",
"Add some bushes and trees in the background."
] | 1,024 | 1,024 | 10 | 9 | The bushes have been successfully removed in the edited image, which matches the editing instruction perfectly. The scene remains largely similar to the original, with minimal changes apart from the removal of the bushes. | 9 | The image has minor distortions, such as the unnatural shape of the birdhouse and the slightly blurred background. However, there are no significant AI-artifacts or unusual body proportions noted. | 9 |
||
task_obj_add_370101 | addition | [
"Add the wall mural",
"Add a stripe pattern on the wall art in the background."
] | 1,152 | 768 | 10 | 9 | The wall mural has been successfully removed in the edited image. The rest of the scene remains almost identical to the original with minimal changes. | 9 | The image has a high level of technical quality with minimal distortions and no obvious unusual body parts or proportions. The shapes of objects are natural, and the overall composition is harmonious. There is minor blurring around the edges of some objects. | 9 |
||
task_obj_add_112528 | addition | [
"Add ceiling light",
"Add a modern black two-armed wall lamp above the sofa."
] | 1,344 | 768 | 10 | 9 | The ceiling light has been successfully removed from the edited image. The overall scene remains highly recognizable with minimal changes to other elements. | 9 | The image appears to be of high quality with minimal distortions. The proportions and shapes of objects such as the furniture are natural, and there are no visible artifacts or unusual body parts. The only minor issue is a slight blur on some edges, but it does not detract significantly from the overall image. | 9 |
||
task_obj_add_545853 | addition | [
"Add the vinyl record",
"Add a black vinyl record with colorful electric sparks coming out of its right side in the center."
] | 1,344 | 768 | 10 | 9 | The vinyl record has been completely removed as per the instruction. The background and surrounding effects have been minimally altered to cover up the area where the record was, maintaining a high degree of similarity with the original image. | 9 | The image is nearly artifact-free with well-defined shapes and proportions. However, there are slight distortions in the scattered pieces of the object that may suggest AI generation. | 9 |
||
task_obj_add_173269 | addition | [
"Add man",
"Add a man in a light tan suit and white shirt, holding a folded piece of paper in his left pocket, walking in the foreground."
] | 1,344 | 768 | 10 | 9 | The man has been successfully removed from the image and the background has been filled in appropriately. The edited image looks very similar to the original with minimal changes beyond the removal of the man. | 9 | The image is largely free from distortions, with natural body proportions and object shapes. The face and other details are clear and well-defined. Minor issues might be present but are not immediately noticeable. | 9 |
||
task_obj_add_905872 | addition | [
"Add car",
"Add a dark red four-door sedan parked in the foreground."
] | 1,152 | 864 | 10 | 9 | The car has been completely removed from the scene as instructed. The background where the car was located is consistent with minimal changes to other elements in the image. | 9 | The image appears to be of high quality with no significant AI-artifacts. The objects and surroundings are well-formed, with natural proportions and shapes. There is a slight blur in the distant background which may indicate minor distortions. | 9 |
||
task_obj_add_845636 | addition | [
"Add standing person",
"Add a man with short brown hair wearing a long-sleeved red and white checkered shirt in the foreground, gesturing with his hands."
] | 1,152 | 864 | 10 | 9 | The standing person has been successfully removed from the scene in the edited image. The rest of the elements and composition remain mostly unchanged, with minimal overediting. | 9 | The image appears to be of high quality with minimal distortions or unusual body parts. The objects and human figures in the image look natural, and there are no significant anomalies. Minor artifacts may exist but aren't immediately noticeable. | 9 |
||
task_obj_add_860981 | addition | [
"Add pole",
"Add a USA Today newspaper vending machine on the corner of the sidewalk."
] | 864 | 1,152 | 10 | 9 | The pole has been completely removed from the scene as per the instruction. The edited image retains most of the original details with minimal changes other than the removal of the pole. | 9 | The image is mostly artifact-free with natural object shapes and proportions. The building, street, mailbox, and traffic lights appear normal and well-formed. However, the floating yellow traffic light on the left side shows some slight distortion, which prevents a perfect score. | 9 |
||
task_obj_add_385998 | addition | [
"Add ghosts",
"Add two children wearing ghost costumes and holding a pumpkin bucket trick-or-treating in front of the gate."
] | 1,152 | 768 | 10 | 9 | The ghosts have been completely removed from the scene. The rest of the image remains largely unchanged, with minimal signs of overediting. | 9 | The image is well-rendered with minimal distortions. The proportions and shapes of objects, including the building and gate, appear natural and consistent. There are no blurred areas or harmonization issues between subjects. Some minor artifacts can be noticed on closer inspection, such as slight texture inconsistencies in certain areas. | 9 |
||
task_obj_add_51988 | addition | [
"Add the suit coat",
"Add silver and gold floral embellishments to the woman's gray dress."
] | 1,152 | 768 | 10 | 9 | The suit coat has been effectively removed from the individual on the left in the edited image. The degree of overediting is minimal as only the suit coat was altered while other elements remain unchanged. | 9 | The image is largely free of significant AI artifacts. The proportions and body parts appear natural, object shapes are well-defined, and faces are clear without blur. There is a minor issue with slight harmonization between subjects, but it's barely noticeable. | 9 |
||
task_obj_add_487954 | addition | [
"Add small canoe",
"Add a small wooden bridge over the stream in the foreground."
] | 1,152 | 768 | 9 | 10 | The small canoe has been successfully removed from the edited image. The rest of the scene remains identical to the original with minimal signs of overediting. | 9 | The image is visually striking with minimal distortions. The waterfall, vegetation, and water surface appear natural and well-rendered. Slight artifacts are present in the finer details of some plants and rocks upon very close inspection. | 9 |
||
task_obj_add_804979 | addition | [
"Add chimney",
"Add a small brick chimney on the roof of the house."
] | 864 | 1,152 | 10 | 9 | The chimney has been successfully removed from the house in the edited image. The surrounding area where the chimney was located looks slightly altered but still consistent with the rest of the building structure. | 9 | The image is mostly artifact-free with well-defined details and natural proportions. The building structure, window shapes, and other architectural elements appear realistic. However, there are minor issues such as slight distortions in the window reflections and some blurriness on the edges of the fence. | 9 |
||
task_obj_add_113226 | addition | [
"Add plant",
"Add a large green plant in a tall white vase on the small table in the background."
] | 1,152 | 768 | 10 | 9 | The plant has been successfully removed from the scene. The rest of the image remains almost unchanged, with minimal overediting. | 9 | The image is of high quality with well-defined objects and no significant distortions or unusual shapes. The proportions are natural, and there are no blurred areas. Slightly noticeable discrepancies in the texture on the sofa fabric might be present but do not detract significantly from the overall quality. | 9 |
||
task_obj_add_416729 | addition | [
"Add the giant",
"Add a giant stone golem in the background."
] | 1,152 | 768 | 10 | 9 | The giant has been successfully removed from the scene. The background and other elements remain largely unchanged except for minor lighting adjustments. | 9 | The image is highly detailed with minimal distortions. The proportions of the rider and horse appear natural, and there are no noticeable unusual object shapes. The sky and stars look well-integrated without any visible blurring or artifacts. | 9 |
||
task_obj_add_454660 | addition | [
"Add the chairs",
"Add tall upholstered dining chairs with button-tufted backs around the table."
] | 1,344 | 768 | 10 | 9 | The chairs have been successfully removed in the edited image. The rest of the scene remains largely unchanged, with minimal additional alterations to the environment. | 9 | The image is highly detailed and well-rendered with minimal distortions. The objects, such as the furniture and decorations, have natural shapes and proportions. However, there are minor inconsistencies in the lighting and shadow effects that suggest slight AI artifacts. | 9 |
||
task_obj_add_390757 | addition | [
"Add earrings",
"Add a small, silver earring to the person's visible ear."
] | 1,344 | 768 | 10 | 9 | The earrings have been successfully removed in the edited image. The rest of the image remains almost identical to the original, indicating minimal overediting. | 9 | The image is of high quality with minimal AI-artifacts. The proportions and body parts appear natural, the face is clear and well-defined, and there are no significant distortions or unusual object shapes. Minor issues may exist but are not immediately noticeable. | 9 |
||
task_obj_add_245077 | addition | [
"Add clipboard",
"Add a stack of papers held in the woman's arm in the foreground."
] | 1,344 | 768 | 10 | 9 | The clipboard has been successfully removed from the edited image. The scene looks very similar to the original with minimal changes apart from the removal of the clipboard. | 9 | The image has a high level of detail with no noticeable distortions, unusual body parts or proportions, or unnatural object shapes. The face is clear and not blurred. The subjects appear harmonized within the scene. Minor artifacts may be present but are not immediately apparent. | 9 |
||
task_obj_add_246745 | addition | [
"Add beige coat",
"Add a light beige blazer over the woman's shirt."
] | 1,152 | 768 | 10 | 9 | The beige coat has been successfully removed in the edited image. The scene remains largely unchanged except for the removal of the coat. | 9 | The image is well-rendered with no noticeable distortions, unusual body parts or proportions, and unnatural object shapes. The face is clear, and the subject is harmonized with the background. Minor artifacts might be present but are not easily detectable. | 9 |
||
task_obj_add_212794 | addition | [
"Add pendant lights",
"Add two geometric, black metal pendant lights hanging from the ceiling."
] | 1,344 | 768 | 9 | 10 | The pendant lights have been successfully removed in the edited image. The rest of the scene remains almost identical to the original image. | 9 | The image shows high technical quality with well-defined objects and natural proportions. There are no noticeable distortions or unusual object shapes. The only minor issue is a slight unnatural blending in the reflections on some surfaces, but it does not significantly detract from the overall image quality. | 9 |
||
task_obj_add_160299 | addition | [
"Add windshield",
"Add a beige steering wheel in the car."
] | 1,344 | 768 | 10 | 9 | The windshield has been successfully removed in the edited image. The rest of the car remains largely unchanged except for some minor modifications around the edges of the roof. | 9 | The image is nearly artifact-free with well-rendered details and proportions. The car's shape looks natural, and the reflections are clear and accurate. There is a minor distortion in the front grille area which slightly affects the overall quality. | 9 |
||
task_obj_add_917633 | addition | [
"Add guitar",
"Add a guitar neck and headstock slightly visible from behind the woman with the microphone headset."
] | 1,152 | 768 | 9 | 10 | The guitar has been successfully removed from the edited image. The scene in the edited image follows the instruction very well with minimal changes to other elements. | 9 | The image shows a high level of detail and realism with minimal distortions. Body proportions and object shapes appear natural. There is no noticeable blurring or harmonization issues among subjects. | 9 |
||
task_obj_add_223501 | addition | [
"Add large plant",
"Add a large green leafy plant in the corner of the room in the background."
] | 1,152 | 768 | 10 | 10 | The large plant has been successfully removed from the edited image without making any other noticeable changes. | 9 | The image is largely free of distortions, unusual body parts or proportions, and unnatural object shapes. The furniture and decor appear natural and well-proportioned. Minor artifacts may be present upon close inspection, but overall the quality is high. | 9.486833 |
||
task_obj_add_737687 | addition | [
"Add gloves",
"Add a pair of purple gloves on the lab counter next to the glass container."
] | 864 | 1,152 | 10 | 9 | The gloves have been successfully removed in the edited image. The overall scene remains largely unchanged, with minimal changes beyond the removal of gloves. | 9 | The image shows minimal distortions and the objects appear natural. There are no unusual body parts or proportions, as there are no humans in the image. The object shapes look correct and the overall scene is coherent. Slight blur on some objects could be observed but it does not significantly affect the quality. | 9 |
||
task_obj_add_274218 | addition | [
"Add bridge",
"Add a stone bridge spanning across the river in the background."
] | 1,344 | 768 | 10 | 9 | The bridge has been successfully removed in the edited image. The scene is still recognizable with minimal changes besides the removal of the bridge. | 9 | The image has high technical quality with minimal distortions. The landscape appears natural, and the objects are well-formed without noticeable unusual shapes or proportions. There is a minor blur in some distant objects, but it does not significantly detract from the overall quality. | 9 |
||
task_obj_add_171334 | addition | [
"Add stream",
"Add a reflection of the trees and sky in the stream in the foreground."
] | 1,344 | 768 | 9 | 9 | The stream has been effectively removed and replaced with land, fulfilling the editing instruction. The overall scene remains largely unchanged except for the removal of the stream. | 9 | The image is highly detailed with vibrant colors and well-defined shapes. There are no noticeable distortions, unusual body parts or proportions, nor unnatural object shapes. The only minor issue could be slight blurring in the distant background, but it does not detract significantly from the overall quality. | 9 |
||
task_obj_add_658945 | addition | [
"Add person",
"Add a person walking on the path in the background."
] | 1,152 | 768 | 10 | 10 | The person has been completely removed from the edited image without affecting other elements. | 9 | The image is nearly artifact-free with natural object shapes and proportions. There are no noticeable distortions or unusual body parts present. However, there is a slight unnatural blending of some grass blades and soil which prevents it from being perfect. | 9.486833 |
||
task_obj_add_143368 | addition | [
"Add bridge",
"Add a bridge in the background."
] | 1,344 | 768 | 10 | 9 | The bridge has been completely removed from the background, following the editing instruction perfectly. The scene retains its original composition with minimal overediting. | 9 | The image is highly realistic with minimal distortions. The car and background elements are well-rendered, but there is a slight blur on the train and minor inconsistencies in lighting. | 9 |
||
task_obj_add_783547 | addition | [
"Add license plate",
"Add a license plate that reads \"CDD 722K\" to the front of the white car."
] | 1,152 | 768 | 10 | 9 | The license plate has been completely removed in the edited image, achieving the editing instruction perfectly. The rest of the car and background remain consistent with minimal overediting. | 9 | The image is of high quality with minimal distortions. The proportions and shapes of the car are natural, and there are no noticeable artifacts or unusual body parts. The only minor issue could be slight blurring in distant objects. | 9 |
||
task_obj_add_375117 | addition | [
"Add black hat",
"Add a black fedora hat on the woman's head."
] | 1,152 | 768 | 10 | 9 | The black hat has been successfully removed in the edited image. The rest of the scene remains largely unchanged with minimal signs of overediting. | 9 | The image has high technical quality with well-defined features and minimal distortions. There are no unusual body parts or proportions, and the object shapes appear natural. The face is clear and not blurred. However, a slight unnatural blending of the background can be noticed upon close inspection. | 9 |
||
task_obj_add_880569 | addition | [
"Add green tree",
"Add a lighted Christmas tree in the background near the water's edge."
] | 1,152 | 768 | 10 | 10 | The green tree has been completely removed in the edited image without any noticeable overediting. | 9 | The image displays a high level of detail and realism with minimal distortions. The proportions and shapes of the objects appear natural, and there are no significant AI-artifacts visible. Slightly noticeable artifacts in the water reflection but overall very effective. | 9.486833 |
||
task_obj_add_506496 | addition | [
"Add the coat",
"Add a dark coat over the person's outfit."
] | 1,344 | 768 | 10 | 9 | The coat has been successfully removed in the edited image. The background and other elements remain consistent with minimal overediting. | 9 | The image has high technical quality with no noticeable distortions, unusual body parts or proportions, and all objects appear natural. The face is clear without any blurring. The only minor issue might be a slight unnatural blending of the background water. | 9 |
||
task_obj_add_439993 | addition | [
"Add person with umbrella",
"Add a person riding a bicycle in the background, holding an umbrella."
] | 1,344 | 768 | 10 | 9 | The person with the umbrella has been completely removed from the scene. The edited image maintains most of the original elements without noticeable overediting. | 9 | The image is nearly artifact-free with well-rendered objects and natural proportions. The bicycles and background elements appear natural, with minimal distortions. There are slight issues in the blending of light and shadow effects, but these are minor. | 9 |
||
task_obj_add_420767 | addition | [
"Add the hair",
"Add gray hair to the man's head."
] | 1,152 | 768 | 9 | 10 | The edited image successfully removed the hair as per the instruction. The scene remains almost identical to the original with minimal overediting. | 9 | The image shows a high level of detail and realism with very minimal distortions. The proportions and body parts appear natural, and there are no noticeable unusual object shapes. The face is clear without any blur or harmonization issues. | 9 |
||
task_obj_add_99943 | addition | [
"Add bar stools",
"Add two modern black bar stools in front of the kitchen island."
] | 1,344 | 768 | 10 | 9 | The bar stools have been successfully removed from the image, following the editing instruction perfectly. The edited image retains most of the original elements and looks minimally altered otherwise. | 9 | The image is highly realistic with no noticeable distortions, unusual body parts or proportions (since there are no humans), and objects have natural shapes. The lighting and shadows look consistent, and the overall composition looks harmonized. However, a slight blur on some distant objects can be observed, preventing a perfect score. | 9 |
||
task_obj_add_113630 | addition | [
"Add office chair",
"Add a woman wearing a light gray jacket and black pants walking in the office."
] | 1,344 | 768 | 10 | 9 | The office chair has been successfully removed from the image. The overall scene remains largely unchanged with minimal differences beyond the removal of the chair. | 9 | The image is highly detailed with well-defined objects and subjects. There are no obvious distortions or unusual body parts/proportions. Object shapes appear natural, and there is good harmony among the subjects. However, a slight blur on some faces can be noticed upon close inspection. | 9 |
||
task_obj_add_856565 | addition | [
"Add camera mount",
"Add a black camera slider with a black camera and a black controller box attached."
] | 1,152 | 864 | 10 | 9 | The camera mount has been successfully removed from the image. The rest of the image remains mostly unchanged except for a slight shadow or outline where the mount was removed. | 9 | The image is largely free of noticeable AI artifacts. The shapes and proportions of the object appear natural, and there are no visible distortions or unusual body parts. However, the shadow could be slightly more refined. | 9 |
||
task_obj_add_919649 | addition | [
"Add man",
"Add a dark-skinned man with dreadlocks, wearing a yellow shirt, black pants, and sunglasses, leaning against the garage door with his arms crossed."
] | 1,152 | 768 | 10 | 9 | The man has been completely removed from the image as instructed. The background and surroundings remain mostly intact with minimal overediting. | 9 | The image has a high level of detail and minimal distortions. The shapes and proportions appear natural, and there are no significant artifacts visible. However, slight inconsistencies in the alignment of the bricks can be noticed on close inspection. | 9 |
||
task_obj_add_591128 | addition | [
"Add license plate",
"Add a C892 NMB license plate to the front of the car."
] | 1,152 | 768 | 10 | 9 | The license plate has been successfully removed in the edited image. The rest of the scene remains largely unchanged, indicating minimal overediting. | 9 | The image appears to be of high quality with no significant AI artifacts. The car and people are well-proportioned, and the objects look natural. However, slight distortions can be noticed in the finer details of the car's body. | 9 |
||
task_obj_add_147298 | addition | [
"Add right shark",
"Add two grey reef sharks swimming in the background."
] | 1,152 | 768 | 10 | 9 | The right shark has been successfully removed as per the editing instruction. The edited image maintains most of the original scene with minimal overediting. | 9 | The image is of high quality with no significant distortions, unusual body parts or proportions, unnatural object shapes, or blurred faces. The subjects are harmonized well within the scene. | 9 |
||
task_obj_add_204068 | addition | [
"Add man's hair",
"Add a child with short black hair, wearing a denim jacket, hugging the man."
] | 1,344 | 768 | 10 | 9 | The man's hair has been successfully removed as per the instruction. The overall scene remains similar with minimal changes outside of the hair removal. | 9 | The image appears to be well-rendered with minimal distortions. The proportions of the body and facial features are natural, and there are no noticeable blurred faces or significant artifacts in object shapes. The background is harmonized well with the subject. | 9 |
||
task_obj_add_395070 | addition | [
"Add the plate of food",
"Add a plate with food being held by the person on the left."
] | 1,152 | 768 | 9 | 10 | The plate of food has been successfully removed from the edited image. The rest of the scene remains unchanged and minimal editing is evident. | 9 | The image shows a high level of detail and realism with minimal noticeable distortions. The body proportions and object shapes appear natural, though there is a slight blur on the faces that could be improved. | 9 |
||
task_obj_add_169358 | addition | [
"Add naan bread",
"Add a paratha on a plate to the background."
] | 1,152 | 768 | 10 | 9 | The naan bread has been successfully removed in the edited image. The overall scene remains very similar to the original with minimal changes except for the removal of the naan bread. | 9 | The image is nearly flawless with realistic textures and proportions. There are no significant distortions, unusual body parts or object shapes, or blurred areas visible. The only minor issue is a slight unnatural blending of colors in some areas. | 9 |
||
task_obj_add_148095 | addition | [
"Add pink dress",
"Add a vibrant pink sari to the woman in the foreground."
] | 1,344 | 768 | 10 | 9 | The pink dress has been completely removed from the image as per the instruction. The rest of the scene remains largely unchanged except for the removal of the person and slight modification in shadows. | 9 | The image is very well-executed with sharp details and proper proportions for both the person and the architectural elements. The lighting and shadows are natural, and there are no apparent distortions or unusual body parts. However, a closer inspection reveals minor inconsistencies in the texture patterns on the archway. | 9 |
||
task_obj_add_513076 | addition | [
"Add the person working",
"Add a man wearing a white tank top and jeans, standing in the garage behind the car, looking at the engine."
] | 1,344 | 768 | 10 | 9 | The person working on the car has been successfully removed from the scene. The background and other elements remain mostly intact with minimal noticeable changes, indicating a high degree of success in both editing execution and preservation. | 9 | The image is nearly artifact-free with well-proportioned and natural-looking elements. The only minor issue is a slight unnatural look in the reflections on the car, but it does not detract significantly from the overall quality. | 9 |
||
task_obj_add_125414 | addition | [
"Add girl",
"Add two anime characters, a boy and girl facing each other with the boy reaching out to the girl, in the foreground."
] | 1,344 | 768 | 10 | 9 | The girl has been completely removed from the edited image, fulfilling the editing instruction perfectly. The background where she stood is filled in effectively to blend with the overall scene. However, there are slight changes in the foreground which makes it slightly less minimal. | 9 | The image is technically well-executed with minimal distortions. The sky and landscape are depicted beautifully, though there's a slight blending issue at the edges of some objects. Overall, it’s close to artifact-free. | 9 |
||
task_obj_add_467818 | addition | [
"Add red flowers",
"Add a variety of colorful wildflowers in the foreground."
] | 1,152 | 768 | 10 | 9 | The red flowers have been successfully removed and replaced with yellow flowers. The overall scene remains very similar to the original image with minimal overediting. | 10 | The image shows no visible distortions, unusual body parts or proportions, unnatural object shapes, blurred faces, or lack of harmony among subjects. The scene appears coherent and artifact-free. | 9.486833 |
||
task_obj_add_756838 | addition | [
"Add V-neck shirt",
"Add a dark blue sweater on the man in the foreground."
] | 1,152 | 768 | 10 | 9 | The V-neck shirt has been successfully removed in the edited image. The scene remains mostly unchanged except for the removal of the shirt, indicating minimal overediting. | 9 | The image appears to be mostly artifact-free with clear and well-proportioned human figures. There is a slight blur on the faces, but it does not detract significantly from the overall quality. | 9 |
||
task_obj_add_70264 | addition | [
"Add green shirt",
"Add a Projects Abroad Fiji logo to the woman's green t-shirt."
] | 1,344 | 768 | 10 | 9 | The green shirt has been successfully removed and replaced with a different clothing item in the edited image. The changes are minimal yet effective. | 9 | The image has a high level of detail and realism. The proportions of the humans appear natural, and there are no significant distortions or unusual body parts. The objects in the background have proper shapes and textures. There is slight blur on the faces, but it's minimal. | 9 |
||
task_obj_add_334592 | addition | [
"Add the computer",
"Add a laptop with a blue screen on the desk."
] | 1,344 | 768 | 10 | 9 | The computer has been successfully removed from the image. The rest of the scene remains largely unchanged except for minor adjustments to the objects on the desk where the computer was previously located. | 9 | The image is well-rendered with minimal noticeable artifacts. The objects and proportions appear natural, and there are no significant distortions or unusual shapes. Only very minor anomalies might be present but they do not detract significantly from the overall quality. | 9 |
||
task_obj_add_355694 | addition | [
"Add the person painting",
"Add a woman with long light brown hair, wearing a black tank top and black leggings, sitting on a chair in front of the painting and painting on it."
] | 1,344 | 768 | 10 | 9 | The person painting has been completely removed in the edited image. The background and other elements remain mostly intact, with minimal changes to accommodate the removal of the person. | 9 | The image is nearly artifact-free with realistic textures and colors. The painting on the wall looks natural, and the objects in the room are well-proportioned and properly shaped. There is a slight unnatural blending where the painting meets the wall frame, but it's minimal. | 9 |
||
task_obj_add_393735 | addition | [
"Add necklace",
"Add a colorful, layered necklace around the person's neck."
] | 1,152 | 768 | 10 | 9 | The necklace has been successfully removed in the edited image. The rest of the scene remains nearly identical to the original with minimal changes. | 9 | The image appears to be of high quality with well-rendered details and natural-looking body proportions. The face is clear and free from blurring. Hair and other features look realistic, although there may be very minor distortions upon extremely close inspection. | 9 |
||
task_obj_add_468446 | addition | [
"Add the glasses",
"Add glasses to the boy in the center wearing a plaid shirt."
] | 1,344 | 768 | 10 | 9 | The glasses have been successfully removed from all children in the edited image. The overall scene remains largely unchanged except for minor adjustments to clothing and background elements. | 9 | The image is highly realistic with well-proportioned body parts and natural object shapes. There are no noticeable distortions or significant AI-artifacts. | 9 |
||
task_obj_add_171343 | addition | [
"Add flowers",
"Add a bouquet of colorful flowers in a white vase on the kitchen island."
] | 1,344 | 768 | 10 | 9 | The flowers have been successfully removed from the kitchen island as per the instruction. The overall scene remains largely unchanged with minimal overediting. | 9 | The image exhibits high technical quality with well-rendered objects and no noticeable distortions or unusual shapes. The proportions of the objects appear natural, and there are no blurred faces as there are no humans present in the scene. The lighting and shadows are harmonious throughout the image. Minor artifacts may be present but not immediately apparent. | 9 |
||
task_obj_add_454052 | addition | [
"Add the face",
"Add a man with a serious expression and slight facial scars in the foreground, wearing a plaid suit and tie."
] | 1,344 | 768 | 10 | 9 | The face has been successfully removed and replaced with a mannequin head, fulfilling the editing instruction perfectly. The edited image retains much of the original scene's clothing and background, indicating minimal overediting. | 9 | The image is nearly artifact-free with well-defined details. The suit and mannequin are depicted realistically without significant distortions or unusual proportions. The only minor issue is a slight blur on the edges of the mannequin's neck area. | 9 |
||
task_obj_add_76471 | addition | [
"Add blue base",
"Add ornate gold details around the base and center of the trophy."
] | 1,152 | 768 | 10 | 9 | The blue base has been successfully removed in the edited image. The rest of the image remains almost identical to the original, indicating minimal overediting. | 9 | The image has a high level of detail and clarity. The proportions and shapes appear natural, and the object is harmonized with the background. However, there is a slight blur in the background which might indicate minor AI artifacts. | 9 |
||
task_obj_add_685716 | addition | [
"Add person avatar",
"Add a female avatar wearing a short-sleeved blue and white checkered shirt and a knee-length kilt to the TV screen."
] | 1,152 | 864 | 10 | 9 | The person avatar has been completely removed from the screen in the edited image. The rest of the scene remains largely unchanged with minimal differences apart from some minor lighting and background adjustments. | 9 | The image is mostly artifact-free with clear details. The objects and proportions appear natural, but there are minor distortions in the texture of some furniture pieces that slightly detract from realism. | 9 |
||
task_obj_add_36159 | addition | [
"Add black suit",
"Add a title card at the bottom that reads: \"WHY IT'S IMPORTANT TO SET UP A TRUST\" in white text with a gray and dark red background, along with a logo on the bottom left corner."
] | 1,344 | 768 | 10 | 9 | The black suit has been successfully removed from the individual in the middle of the image. The editing is minimal and effective as it does not significantly alter other elements of the scene. | 9 | The image is mostly free of distortions, with no unusual body parts or proportions. The objects in the background appear natural and are well-defined. There is a minor blur on one face which slightly affects the overall quality. | 9 |
||
task_obj_add_484061 | addition | [
"Add the roof",
"Add a large wooden barn in the foreground of the image."
] | 1,344 | 768 | 10 | 9 | The roof of the building has been successfully removed as per the instruction. The background and landscape have remained very similar to the original image with minimal changes. | 10 | The image shows no noticeable distortions, unusual body parts or proportions (as it is a landscape), unnatural object shapes, blurred faces, or any disharmonized subjects. The image appears artifact-free. | 9.486833 |
||
task_obj_add_161634 | addition | [
"Add flower bouquet",
"Add a bouquet of sunflowers in the foreground being held by the man."
] | 1,344 | 768 | 10 | 9 | The flower bouquet has been successfully removed in the edited image. The changes are minimal and only affect the area where the bouquet was originally present. | 9 | The image is nearly artifact-free with very well-defined features. The facial details are clear, and the proportions of the body parts appear natural. However, there is a minor issue with the hand holding the medal which looks slightly unnatural in terms of position. | 9 |
||
task_obj_add_74966 | addition | [
"Add iPad",
"Add a tablet in the man's hands."
] | 1,344 | 768 | 10 | 9 | The iPad has been completely removed in the edited image. The rest of the scene remains largely unchanged with minimal alterations. | 9 | The image is nearly artifact-free with well-proportioned human features and natural-looking object shapes. Slight blurring of the background can be seen, but it appears intentional to focus on the subject. | 9 |
||
task_obj_add_553968 | addition | [
"Add dark background",
"Add a dark, textured background behind the beer bottle."
] | 1,024 | 1,024 | 9 | 10 | The dark background has been successfully removed and replaced with a brighter background. The edited image maintains the original composition without any noticeable overediting. | 9 | The image is mostly clear and well-rendered with no significant distortions or unusual object shapes. The text is legible, though slightly blurred in some areas. | 9 |
||
task_obj_add_558201 | addition | [
"Add the person listening",
"Add a woman and a man in U.S. Air Force uniforms in front of the whiteboard."
] | 1,152 | 768 | 10 | 9 | The person listening has been successfully removed from the scene. The background where the person stood has been filled seamlessly with minimal disturbance to the original image. | 9 | The image is largely free of noticeable distortions, unusual body parts or proportions, and unnatural object shapes. The text on the board is clear and legible, and the colors are well-rendered with no obvious blurriness. Minor artifacts may be present but they do not detract significantly from the overall quality. | 9 |
||
task_obj_add_865362 | addition | [
"Add books",
"Add several colorful books falling into the funnel on the person's head."
] | 1,152 | 864 | 10 | 9 | The books have been successfully removed from the character's head in the edited image. The rest of the image remains largely unchanged with minimal alterations. | 9 | The image is well-rendered with no significant distortions, unusual body parts or proportions, and unnatural object shapes. The subject is harmonized, and there are no blurred faces. The only minor issue is slight stylization that could be intentional. | 9 |
||
task_obj_add_817610 | addition | [
"Add broken windows",
"Add scaffolding inside one of the archways."
] | 1,152 | 864 | 10 | 9 | The broken windows have been successfully removed in the edited image. The scene remains largely unchanged except for the removal of window panes and slight color adjustments around the area where the windows were removed. | 9 | The image has a high level of detail and realism. The proportions and shapes of the structure appear natural. There are no noticeable distortions or unusual object shapes. However, there is a minor artifact near the lower left corner where the texture seems slightly off. | 9 |
||
task_obj_add_404707 | addition | [
"Add the girl in yellow",
"Add a girl with a backwards baseball cap and a yellow and black crop top on a bike behind the boy in the foreground."
] | 1,344 | 768 | 10 | 9 | The girl in yellow has been successfully removed from the edited image. The background and other elements remain largely unchanged, with minimal signs of overediting. | 9 | The image is nearly artifact-free with well-proportioned body parts and natural object shapes. There are no significant distortions or unusual proportions. Minor discrepancies in the background, but they are negligible. | 9 |
||
task_obj_add_111791 | addition | [
"Add right owl",
"Add three small brown owls, two perched together on the wooden post and one standing slightly apart on the wooden beam."
] | 1,152 | 768 | 10 | 9 | The right owl has been successfully removed from the image. The edited area is seamlessly integrated into the background, and there are minimal signs of overediting. | 9 | The image is largely artifact-free with natural-looking textures and shapes. The only minor issue could be slight blurring in the background, which might not be intentional but can also be a stylistic choice. | 9 |
||
task_obj_add_910708 | addition | [
"Add red wagon",
"Add a small red and gray wagon in the background."
] | 1,152 | 768 | 10 | 9 | The red wagon has been successfully removed from the scene. The overall composition and elements of the image remain consistent with minimal changes beyond the removal of the wagon. | 9 | The image is generally well-rendered with minimal distortions. The body proportions and object shapes appear natural. There are minor anomalies in the details of some faces, but they do not significantly detract from the overall quality. | 9 |
||
task_obj_add_199457 | addition | [
"Add trench coat",
"Add a light brown trench coat over the man's suit."
] | 1,344 | 768 | 10 | 9 | The trench coat has been successfully removed and replaced with a suit. The rest of the image remains very similar to the original, indicating minimal overediting. | 9 | The image shows high technical quality with well-defined facial features and proper proportions. There is a slight distortion in the background, but it does not detract significantly from the overall effectiveness of the image. | 9 |
||
task_obj_add_827982 | addition | [
"Add black tower",
"Add a tall, oval-shaped skyscraper in the background."
] | 1,152 | 768 | 10 | 9 | The black tower has been successfully removed in the edited image. The overall scene remains consistent with minimal changes aside from the removal of the tower. | 9 | The image exhibits minimal distortions and the buildings have natural shapes. The proportions of the buildings appear accurate, and there are no noticeable unusual body parts or significant AI-artifacts. There is a slight haze in some areas which could be an atmospheric effect rather than an artifact. | 9 |
||
task_obj_add_87915 | addition | [
"Add tennis racket",
"Add a black and gray tennis racket in the foreground."
] | 1,344 | 768 | 9 | 10 | The tennis racket has been successfully removed from the edited image. The rest of the image remains largely unchanged and recognizable. | 9 | The image has high technical quality with well-defined features. There are no significant distortions or unusual body parts visible. The object shapes appear natural, and there is a clear focus on the face without any noticeable blurring. However, there might be minor imperfections that could be overlooked at first glance. | 9 |
||
task_obj_add_514651 | addition | [
"Add the jacket",
"Add a dark brown suede jacket on the man."
] | 1,344 | 768 | 10 | 9 | The jacket has been successfully removed in the edited image. The overall scene and background remain largely unchanged, indicating minimal overediting. | 9 | The image appears to be very well-rendered with minimal distortions. The proportions and body parts look natural, and the face is clear without any blurring. However, there might be slight unnaturalness in the background blur, but it's minor. | 9 |
||
task_obj_add_406262 | addition | [
"Add red boat",
"Add a small, weathered wooden rowboat floating in the water near the dock."
] | 1,152 | 768 | 10 | 9 | The red boat has been completely removed from the edited image. The rest of the scene remains largely unchanged, maintaining the integrity of the original image. | 9 | The image is highly realistic with excellent detail and reflection. The house, trees, and water are well-rendered with minimal distortions or unusual object shapes. The only minor issue could be the slight unnatural blending of colors in some areas of the foliage. | 9 |
||
task_obj_add_923614 | addition | [
"Add exit signs",
"Add a green highway sign for exits 56A and 56B, indicating directions to Raleigh and Sanford."
] | 1,152 | 768 | 10 | 10 | The exit signs have been successfully removed as per the instruction. The edited image closely resembles the original with minimal changes beyond the removal of exit signs. | 9 | The image appears to be largely artifact-free with well-defined structures and natural-looking objects. However, there is a slight blur or halo around the top of the structure in the center which could indicate minor AI-generated artifacts. | 9.486833 |
||
task_obj_add_244954 | addition | [
"Add pipe",
"Add a brown pipe to the man's mouth."
] | 1,344 | 768 | 10 | 9 | The pipe has been successfully removed from the edited image. The rest of the scene remains largely unchanged, with minimal differences in the background blur. | 9 | The image has high technical quality with no significant distortions, unusual body parts or proportions, or unnatural object shapes. The face is clear and not blurred. There is a minor issue with the alignment of the chair's seatrest, but it does not detract significantly from the overall quality. | 9 |
||
task_obj_add_542573 | addition | [
"Add climber's legs",
"Add a person in a red shirt climbing in the center foreground, with climbing gear visible."
] | 768 | 1,152 | 10 | 9 | The climber's legs have been completely removed in the edited image. The rest of the scene remains mostly unchanged with minimal signs of overediting. | 10 | The image is free of distortions, unusual object shapes, and blurred areas. The subjects appear harmonized. | 9.486833 |
OmniEdit
In this paper, we present OMNI-EDIT, which is an omnipotent editor to handle seven different image editing tasks with any aspect ratio seamlessly. Our contribution is in four folds: (1) OMNI-EDIT is trained by utilizing the supervision from seven different specialist models to ensure task coverage. (2) we utilize importance sampling based on the scores provided by large multimodal models (like GPT-4o) instead of CLIP-score to improve the data quality.
📃Paper | 🌐Website | 💻Github | 📚Dataset
Data Pipeline
We synthesize the large scale dataset through specialist distillation. Our synthesis pipeline is depicted in
Our released version contains 1.2M pairs covering seven different skills like addition, swaping, removal, attribute modification, background change, environment change and sytle transfer. The dataset has been filtered with VIEScore.
Comparison with Others
Our dataset has the most diverse, highest-quality image editing pairs of any resolution.
Citation
If you find our paper useful, please cite us with
@article{wei2024omniedit,
title={OmniEdit: Building Image Editing Generalist Models Through Specialist Supervision},
author={Wei, Cong and Xiong, Zheyang and Ren, Weiming and Du, Xinrun and Zhang, Ge and Chen, Wenhu},
journal={arXiv preprint arXiv:2411.07199},
year={2024}
}
- Downloads last month
- 347