Thats Not Me

2026-02-27 13:25:16 UTC · 1275.6s · 5 models · 9 inputs · Total cost: $1.36
#1 seedream-edit 4.9/5 WINNER
5 models · 9 inputs (n=9) · $1.36 · 1276s · ⚠ Small sample — scores may vary with more prompts

Model Rankings

Rank Model Visual Quality Input Fidelity Identity Preservation face_similarity Overall (weighted) Conf Success
1 seedream-editWINNER 4.9 5.0 4.8 0.939 4.9 n=9 97% 9/9
2 flux-kontext 5.0 4.7 4.9 0.877 4.7 n=9 96% 9/9
3 reve-edit 4.6 4.2 4.3 0.910 4.4 n=9 92% 9/9
4 flux-dev-i2i 4.9 1.0 1.0 0.029 2.3 n=9 98% 9/9
5 firered-edit 0.0 0.0 0.0 0.000 0.0 n=0 100% 0/9

Dimension Profile

Cost Efficiency

RankModelScoreCost/ImageScore/$vs Winner
1 seedream-editWINNER 4.9 $0.0300 163 baseline
2 reve-edit 4.4 $0.0400 110 -0.5 quality, 1x cost
3 flux-kontext 4.7 $0.0500 93 -0.2 quality, 2x cost
4 flux-dev-i2i 2.3 $0.0250 92 -2.6 quality, 17% cheaper
⚠ Metric Warnings

Metric-VLM Correlation

PairPearson rp-valueAgreement
face similarity vs identity preservation 0.91 0.0000 high agreement

Place this person in a professional office with a city view behind them

Input Input image
Reference image
reve-edit reve-edit
5.0/5
18292ms · $0.0400
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits professional quality with excellent sharpness, natural lighting, and coherent rendering of the subject and background. There are no visible artifacts, distortions, or inconsistencies, making it appear like a high-quality photograph.
Input Fidelity: 5.0/5 (100% conf)
The transformed image perfectly preserves the identity, facial features, expression, and clothing of the man from the original image. The only significant change is the background, which is a deliberate transformation rather than a loss of the subject's key features.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved. All facial features, including eye shape, nose, mouth, and jawline, remain identical. The skin tone, complexion, and expression are also fully consistent between the two images.
face_similarity: 0.8110
seedream-edit seedream-edit
5.0/5
31229ms · $0.0300
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits excellent visual quality with sharp focus on the subject, natural lighting, and realistic textures. There are no visible artifacts, distortions, or inconsistencies, making it appear professional and high-fidelity.
Input Fidelity: 5.0/5 (100% conf)
The transformed image perfectly preserves the identity, facial features, expression, and pose of the man from the original input. The only significant change is the background, which does not impact the fidelity of the main subject.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved. All facial features, skin tone, and expression remain identical between the original and transformed images.
face_similarity: 0.9516
flux-kontext flux-kontext
5.0/5
9219ms · $0.0500
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits professional quality with excellent sharpness, realistic lighting, and natural textures across the subject's face, hair, and clothing. There are no visible artifacts, distortions, or inconsistencies, making it indistinguishable from a high-quality photograph.
Input Fidelity: 5.0/5 (95% conf)
The transformed image perfectly preserves the identity of the subject, including his facial features, hairstyle, beard, and even the specific white V-neck shirt. The pose and expression are also maintained with remarkable accuracy, making the subject instantly recognizable despite the background change.
Identity Preservation: 5.0/5 (90% conf)
The individual's identity is perfectly preserved. All key facial features, including eye shape, nose, mouth, and jawline, are consistent. The hair, beard style, and overall expression are also identical.
face_similarity: 0.9700
firered-edit
Failed: Request cf189add-70b5-44c0-9180-b2551ea0d161 timed out after 300 seconds
flux-dev-i2i flux-dev-i2i
2.3/5
2942ms · $0.0250
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits excellent visual quality, appearing professional and highly realistic. There are no visible artifacts, distortions, or inconsistencies, with sharp focus on the subject and natural rendering of details.
Input Fidelity: 1.0/5 (100% conf)
The transformed image features a completely different person (a woman) compared to the original input (a man). All key identifying features such as gender, facial structure, hair, and clothing have been entirely altered, resulting in a complete loss of the original subject's identity.
Identity Preservation: 1.0/5 (100% conf)
The identity has been completely changed. The original image features a man, while the transformed image shows a different woman. All facial features, skin tone, and body proportions are entirely different.
face_similarity: 0.0123
⚠ Face similarity (0.0123) below threshold (0.6) — excluded from overall

Transform this into a warm-toned vintage film photograph

Input Input image
Reference image
reve-edit reve-edit
3.7/5
19997ms · $0.0400
Score Details
Visual Quality: 4.0/5 (85% conf)
The image demonstrates good visual quality with sharp facial features and a clear subject. However, the hair lacks some natural detail and the strong warm color cast, while potentially artistic, might be considered a stylistic choice rather than a neutral rendering.
Input Fidelity: 4.0/5 (90% conf)
The transformed image strongly preserves the man's identity, facial features, expression, and clothing from the original. The composition and background elements are also perfectly maintained, with the only significant difference being a strong yellow/orange color cast applied to the entire output image.
Identity Preservation: 3.0/5 (80% conf)
The person's facial features, expression, and body proportions are perfectly preserved, making them instantly recognizable. However, the skin tone and overall image color balance have shifted significantly, giving the subject a much warmer, more orange complexion compared to the original, which is a noticeable difference.
face_similarity: 0.9532
seedream-edit seedream-edit
4.3/5
31171ms · $0.0300
Score Details
Visual Quality: 5.0/5 (90% conf)
The image exhibits excellent visual quality, characterized by a consistent and well-executed vintage aesthetic with prominent film grain. The subject's features are natural and sharply rendered, and there are no discernible AI generation artifacts or inconsistencies.
Input Fidelity: 5.0/5 (100% conf)
The transformed image perfectly preserves the identity of the subject, his facial features, expression, and clothing. The composition and pose are also identical, with the changes being purely stylistic, involving a warm color shift, added texture, and a vignette.
Identity Preservation: 3.0/5 (90% conf)
The individual's facial features, expression, and body proportions are perfectly preserved, making them instantly recognizable. However, the overall skin tone and image color grading have been significantly altered, shifting from a natural look to a much warmer, sepia-like tone.
face_similarity: 0.9590
flux-kontext flux-kontext
5.0/5
13023ms · $0.0500
Score Details
Visual Quality: 5.0/5 (95% conf)
The image exhibits excellent visual quality with sharp focus, natural lighting, and realistic textures. There are no visible artifacts, distortions, or blurring, making it appear professional and high-fidelity.
Input Fidelity: 5.0/5 (100% conf)
The transformed image exhibits excellent input fidelity, perfectly preserving the subject's identity, facial features, expression, and clothing. The background and overall composition are also identical to the original input image.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved. All facial features, skin tone, and expression remain identical between the original and transformed images.
face_similarity: 0.9379
firered-edit
Failed: Request 44f01611-7ac3-4942-a950-b570c6967098 timed out after 300 seconds
flux-dev-i2i flux-dev-i2i
2.3/5
2368ms · $0.0250
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits excellent visual quality, appearing professional and highly realistic. There are no visible artifacts, distortions, or blurring, and all elements are rendered with exceptional clarity and coherence.
Input Fidelity: 1.0/5 (100% conf)
The transformed image completely replaces the original subject with a different person, altering all key features such as gender, facial characteristics, hair, and clothing. There is no preservation of the original input's identity or appearance.
Identity Preservation: 1.0/5 (100% conf)
The identity of the person in the transformed image is completely different from the original. The original image features a man, while the transformed image shows a woman, indicating a total change in identity.
face_similarity: 0.0360
⚠ Face similarity (0.0360) below threshold (0.6) — excluded from overall

Add dramatic studio lighting with a dark background

Input Input image
Reference image
reve-edit reve-edit
4.3/5
18405ms · $0.0400
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits excellent visual quality, characterized by sharp focus, natural skin texture, and professional lighting. There are no visible artifacts, distortions, or inconsistencies, making it appear highly realistic and polished.
Input Fidelity: 4.0/5 (85% conf)
The output image demonstrates strong preservation of the main subject, including the man's identity, facial features, expression, and clothing. However, the background has been completely altered from a light, textured wall to a dark, plain one, which is a significant change to the overall composition.
Identity Preservation: 4.0/5 (90% conf)
The individual's identity is strongly preserved. Facial features such as eye shape, nose, mouth, and jawline remain consistent, and the expression is identical. The main difference is a change in lighting and background, which subtly alters the perceived skin tone but does not affect recognition.
face_similarity: 0.9597
seedream-edit seedream-edit
5.0/5
39015ms · $0.0300
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits excellent visual quality with professional-grade sharpness, lighting, and detail. There are no visible artifacts, distortions, or inconsistencies typically associated with AI generation.
Input Fidelity: 5.0/5 (100% conf)
The transformed image perfectly preserves the identity, facial features, expression, and pose of the man from the original image. While the background and lighting have been dramatically altered, the core subject remains entirely consistent.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved. All facial features, including eye shape, nose, mouth, and jawline, remain identical. The expression and hairstyle are also fully consistent between the two images.
face_similarity: 0.8584
flux-kontext flux-kontext
5.0/5
9980ms · $0.0500
Score Details
Visual Quality: 5.0/5 (95% conf)
The image exhibits excellent visual quality with professional-grade rendering. There are no visible artifacts, distortions, or blurring, and details are sharp and coherent.
Input Fidelity: 5.0/5 (100% conf)
The transformed image perfectly preserves the subject's identity, facial features, expression, hair, beard, and clothing. The pose and composition of the man are also identical, despite significant changes to the background and lighting style.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved across both images. All facial features, including eye shape, nose, mouth, and jawline, remain identical, with only lighting and background changing.
face_similarity: 0.6576
firered-edit
Failed: Request c8a15399-9447-4efc-845f-b001733d1e58 timed out after 300 seconds
flux-dev-i2i flux-dev-i2i
2.3/5
1822ms · $0.0250
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits excellent visual quality, appearing professionally rendered with sharp focus on the subject's face and natural skin texture. There are no visible artifacts, distortions, or inconsistencies in any elements.
Input Fidelity: 1.0/5 (100% conf)
The transformed image features a completely different person (a woman) compared to the original input image (a man). All key identifying features such as gender, facial structure, hair, and clothing have been lost, indicating no preservation of the original subject's identity.
Identity Preservation: 1.0/5 (100% conf)
The identity of the person has been completely changed. The original image features a man, while the transformed image shows a different woman, making the original person unrecognizable.
face_similarity: 0.0234
⚠ Face similarity (0.0234) below threshold (0.6) — excluded from overall

Place this person in a professional office with a city view behind them

Input Input image
Reference image
reve-edit reve-edit
5.0/5
18142ms · $0.0400
Score Details
Visual Quality: 5.0/5 (95% conf)
The image demonstrates excellent visual quality, appearing sharp, well-lit, and free from any noticeable AI-generated artifacts or distortions. The subject's features, hair, and clothing are rendered with high fidelity and naturalness.
Input Fidelity: 5.0/5 (95% conf)
The output image perfectly preserves the identity, facial features, expression, hair, and pose of the woman from the input. The only noticeable change is the background and a very minor difference in the draping of the lower part of her red garment, which does not detract from the overall fidelity of the key subject.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved across both images. All facial features, including the eyes, nose, mouth, and jawline, remain identical, along with the specific wide, laughing expression.
face_similarity: 0.8657
seedream-edit seedream-edit
5.0/5
31946ms · $0.0300
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits excellent visual quality, appearing professional with no visible artifacts, distortions, or blurring. The subject is sharply in focus, and the overall rendering is clean and realistic.
Input Fidelity: 5.0/5 (100% conf)
The output image demonstrates excellent input fidelity, perfectly preserving the identity, facial expression, pose, and clothing of the woman from the original image. The transformation successfully replaces the background while maintaining every key feature of the subject.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved. All facial features, skin tone, hair, and the exact expression are identical between the original and transformed images.
face_similarity: 0.9613
flux-kontext flux-kontext
4.7/5
8925ms · $0.0500
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits excellent visual quality with no discernible artifacts, distortions, or blurring. The subject is sharply in focus, and details in her face, hair, and clothing are rendered with professional clarity and realism.
Input Fidelity: 4.0/5 (90% conf)
The main subject, the woman, is excellently preserved in terms of her identity, facial features, expression, pose, and clothing. However, the background has been completely replaced, changing from a blurred, indistinct setting to a clear city skyline, which is a significant alteration to the overall scene.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved. All facial features, including eye shape, nose, mouth, and jawline, remain identical. The expression, skin tone, and visible body proportions are also fully consistent between the original and transformed images.
face_similarity: 0.9614
firered-edit
Failed: Request a2b5354d-e151-4671-be8e-f293997139e3 timed out after 300 seconds
flux-dev-i2i flux-dev-i2i
2.3/5
1815ms · $0.0250
Score Details
Visual Quality: 5.0/5 (95% conf)
The image exhibits excellent visual quality with sharp details, natural lighting, and realistic textures. There are no visible artifacts, distortions, or inconsistencies, making it appear professionally rendered.
Input Fidelity: 1.0/5 (100% conf)
The transformed image features a completely different person with distinct facial features, age, hair, and clothing compared to the original input. The background and overall setting have also been entirely changed, indicating a complete loss of key features.
Identity Preservation: 1.0/5 (100% conf)
The transformed image features a completely different individual. All facial features, including eye shape, nose, mouth, and jawline, are altered, making the original person entirely unrecognizable.
face_similarity: 0.0000
⚠ Face similarity (0.0000) below threshold (0.6) — excluded from overall

Transform this into a warm-toned vintage film photograph

Input Input image
Reference image
reve-edit reve-edit
2.3/5
19551ms · $0.0400
Score Details
Visual Quality: 2.0/5 (85% conf)
The image exhibits significant pervasive graininess across the entire frame, which severely detracts from its overall sharpness and clarity. This texture makes the image appear unrefined and compromises the visual fidelity of the subject's skin, hair, and clothing.
Input Fidelity: 2.0/5 (70% conf)
While the woman's identity, pose, and expression are perfectly preserved, the overall color palette has been significantly altered. The vibrant red of her top is changed to a muted reddish-brown, and the entire image has a strong warm, sepia-like tint, significantly deviating from the original's color grading.
Identity Preservation: 3.0/5 (80% conf)
The person's facial features, expression, and body proportions are excellently preserved, making them immediately recognizable. However, the skin tone and overall complexion have been noticeably altered, shifting from a natural tone to a significantly warmer, more desaturated, and sepia-like appearance.
face_similarity: 0.9232
seedream-edit seedream-edit
5.0/5
29087ms · $0.0300
Score Details
Visual Quality: 5.0/5 (90% conf)
The image exhibits excellent visual quality, with a sharp and expressive subject. The intentional film grain and warm color grading contribute to a cohesive and appealing vintage aesthetic, without any noticeable generation artifacts or anatomical inconsistencies.
Input Fidelity: 5.0/5 (95% conf)
The transformed image perfectly preserves the identity, facial features, expression, and pose of the woman from the original input. The changes are limited to a stylistic filter, including a color shift, increased contrast, and a textured border, without altering any core elements of the subject.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved. All facial features, including eye shape, nose, mouth, and jawline, remain identical. The expression, body proportions, and posture are also fully consistent between the original and transformed images.
face_similarity: 0.9472
flux-kontext flux-kontext
5.0/5
9764ms · $0.0500
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits excellent visual quality, appearing professional and natural. There are no visible artifacts, distortions, or inconsistencies typically associated with AI generation, with sharp focus and realistic rendering of all elements.
Input Fidelity: 5.0/5 (100% conf)
The transformed image perfectly preserves all key features from the original input. The subject's identity, facial expression, clothing, hair, and the overall composition and background are identical, showing no discernible losses or alterations.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved. All facial features, skin tone, body proportions, and the joyful expression are identical between the original and transformed images.
face_similarity: 0.9605
firered-edit
Failed: Request 48f7c3a7-e856-4138-8d8c-c77640c35163 timed out after 300 seconds
flux-dev-i2i flux-dev-i2i
2.0/5
2335ms · $0.0250
Score Details
Visual Quality: 4.0/5 (90% conf)
The image exhibits good visual quality with a well-rendered subject, natural skin tones, and realistic hair. However, minor inconsistencies in sharpness and a slight artificiality in some details prevent it from achieving an excellent score.
Input Fidelity: 1.0/5 (95% conf)
The transformed image completely fails to preserve the identity of the original subject. The facial features, hair color and style, and expression are entirely different, making the person unrecognizable. The background has also been completely changed.
Identity Preservation: 1.0/5 (100% conf)
The transformed image features a completely different person. All key facial features, including eye shape and color, nose, mouth, jawline, and hair, have been significantly altered, making the original person unrecognizable.
face_similarity: 0.0393
⚠ Face similarity (0.0393) below threshold (0.6) — excluded from overall

Add dramatic studio lighting with a dark background

Input Input image
Reference image
reve-edit reve-edit
5.0/5
19225ms · $0.0400
Score Details
Visual Quality: 5.0/5 (95% conf)
The image exhibits excellent visual quality, characterized by sharp focus, natural skin texture, and well-defined details in the hair and clothing. The lighting is professional and enhances the subject's features without introducing any artifacts or inconsistencies.
Input Fidelity: 5.0/5 (90% conf)
The output image demonstrates excellent input fidelity, perfectly preserving the identity, facial features, expression, hair, and clothing of the woman from the original image. The transformation primarily involves a change in background and lighting, rather than alteration of the subject herself.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved across both images. All facial features, including eye shape, nose, mouth, and jawline, remain identical, and the expression is fully maintained.
face_similarity: 0.7899
seedream-edit seedream-edit
5.0/5
26136ms · $0.0300
Score Details
Visual Quality: 5.0/5 (95% conf)
The image exhibits excellent visual quality, characterized by sharp focus, natural skin texture, and well-defined features. There are no visible artifacts, blurring, or inconsistencies, contributing to a professional and polished appearance.
Input Fidelity: 5.0/5 (90% conf)
The transformed image exhibits excellent input fidelity, perfectly preserving the identity, facial features, expression, pose, and clothing of the woman from the original image. The only significant alteration is the complete change of the background.
Identity Preservation: 5.0/5 (100% conf)
The identity of the woman is perfectly preserved between the original and transformed images. All facial features, including her eyes, nose, mouth, and jawline, remain identical. Her skin tone, complexion, and the joyful expression are also fully consistent.
face_similarity: 0.9437
flux-kontext flux-kontext
5.0/5
12411ms · $0.0500
Score Details
Visual Quality: 5.0/5 (95% conf)
The image exhibits excellent visual quality, characterized by sharp focus, natural skin texture, and vibrant, well-defined features. There are no visible artifacts, blurring, or inconsistencies, giving it a professional and polished appearance.
Input Fidelity: 5.0/5 (95% conf)
The transformed image perfectly preserves the identity, facial features, expression, hair, and clothing of the woman from the original image. The pose and framing of the subject are also identical, with the only change being the background.
Identity Preservation: 5.0/5 (95% conf)
The identity of the person is perfectly preserved across both images. All facial features, expression, and body proportions remain consistent, with only minor differences attributable to changes in lighting and background.
face_similarity: 0.6529
firered-edit
Failed: Request 1ad1d8ed-740a-4d83-9a7b-1fadd8a555aa timed out after 300 seconds
flux-dev-i2i flux-dev-i2i
2.3/5
2105ms · $0.0250
Score Details
Visual Quality: 5.0/5 (90% conf)
The image exhibits professional-level quality with sharp details, natural lighting, and realistic rendering of the subject's features, hair, and clothing. There are no visible AI artifacts or inconsistencies.
Input Fidelity: 1.0/5 (95% conf)
The transformed image depicts a completely different person with distinct facial features, hair color, and expression. The clothing style and background have also been entirely altered, indicating a complete loss of the original input's identity and context.
Identity Preservation: 1.0/5 (100% conf)
The identity of the person has been completely changed. The transformed image features a different individual with distinct facial features, hair color, and expression.
face_similarity: 0.0000
⚠ Face similarity (0.0000) below threshold (0.6) — excluded from overall

Place this person in a professional office with a city view behind them

Input Input image
Reference image
reve-edit reve-edit
4.3/5
18951ms · $0.0400
Score Details
Visual Quality: 5.0/5 (95% conf)
The image exhibits excellent visual quality with sharp focus on the subject, natural lighting, and no visible artifacts or distortions. Details in the hair, beard, and sweater are rendered with high clarity.
Input Fidelity: 3.0/5 (70% conf)
The main subject's identity, facial features, and hair are perfectly preserved between the input and output images. However, the specific design and color of the sweater are significantly altered, and the entire background has been completely changed from a plain black backdrop to an office setting, representing a noticeable loss of original scene features.
Identity Preservation: 5.0/5 (95% conf)
The identity of the person is perfectly preserved across both images. All key facial features, hair, and beard remain consistent, making the individual instantly recognizable despite changes in lighting and attire.
face_similarity: 0.9502
seedream-edit seedream-edit
5.0/5
38017ms · $0.0300
Score Details
Visual Quality: 5.0/5 (90% conf)
The image exhibits professional quality with exceptional sharpness, realistic textures, and coherent rendering. There are no visible AI artifacts, distortions, or inconsistencies.
Input Fidelity: 5.0/5 (95% conf)
The transformed image perfectly preserves the identity, facial features, hair, beard, and clothing of the man from the original input. The subject's pose and composition are also maintained without any alterations, with only the background being changed.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved. All facial features, skin tone, and body proportions remain consistent between the original and transformed images.
face_similarity: 0.9616
flux-kontext flux-kontext
5.0/5
11069ms · $0.0500
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits excellent visual quality with no discernible artifacts, distortions, or inconsistencies. The subject is sharply focused, and details like hair, skin texture, and the intricate knit of the sweater are rendered with professional clarity and realism.
Input Fidelity: 5.0/5 (95% conf)
The transformed image exhibits excellent input fidelity, perfectly preserving the identity of the man, including his facial features, hair, and beard. The specific details of his grey knitted sweater and the overall pose are also maintained without any alterations.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved across both images. All facial features, including eye shape, nose, mouth, and jawline, remain consistent. Skin tone, complexion, and body proportions are also accurately maintained despite the change in background and lighting.
face_similarity: 0.9208
firered-edit
Failed: Request e5662512-5b74-4d09-ac73-cfec6da35971 timed out after 300 seconds
flux-dev-i2i flux-dev-i2i
2.3/5
1814ms · $0.0250
Score Details
Visual Quality: 5.0/5 (95% conf)
The image exhibits excellent visual quality with sharp focus on the subject and a pleasing, natural aesthetic. There are no visible artifacts, distortions, or inconsistencies, and details like skin texture, hair, and clothing are rendered with high fidelity.
Input Fidelity: 1.0/5 (100% conf)
The transformed image completely fails to preserve the identity and key features of the original subject. The person, clothing, background, and lighting are all entirely different, making it an unrecognizable transformation.
Identity Preservation: 1.0/5 (100% conf)
The identity of the person is not preserved at all. The transformed image features a completely different individual, making the original person unrecognizable. All major facial features have been significantly altered.
face_similarity: 0.0737
⚠ Face similarity (0.0737) below threshold (0.6) — excluded from overall

Transform this into a warm-toned vintage film photograph

Input Input image
Reference image
reve-edit reve-edit
4.7/5
18866ms · $0.0400
Score Details
Visual Quality: 5.0/5 (95% conf)
The image exhibits excellent visual quality, characterized by professional rendering, sharp focus, and intricate details. There are no visible artifacts, blurring, or distortions present.
Input Fidelity: 5.0/5 (95% conf)
The transformed image perfectly preserves the subject's identity, facial features, clothing, and overall composition from the input. The only noticeable difference is a global color shift towards warmer tones in the output image, which does not detract from feature preservation.
Identity Preservation: 4.0/5 (85% conf)
The transformed image maintains a strong resemblance to the original person, with all facial features, hair, beard, and expression perfectly preserved. The primary difference is a noticeable shift in the overall color grading, resulting in a warmer, slightly more yellowish skin tone in the output image compared to the original.
face_similarity: 0.9658
seedream-edit seedream-edit
4.7/5
27239ms · $0.0300
Score Details
Visual Quality: 4.0/5 (90% conf)
The image demonstrates high visual quality with realistic rendering of the subject's features and clothing. However, a pervasive grain/texture filter is applied throughout, which, while potentially an artistic choice, slightly detracts from absolute clarity and sharpness.
Input Fidelity: 5.0/5 (100% conf)
The transformed image perfectly preserves the subject's identity, facial features, clothing details, and overall composition from the original image. The primary change is a stylistic color shift and vintage effect, which does not detract from the preservation of key structural elements.
Identity Preservation: 5.0/5 (95% conf)
The identity of the person is perfectly preserved. All facial features, body proportions, and expression remain identical between the original and transformed images.
face_similarity: 0.9572
flux-kontext flux-kontext
5.0/5
12683ms · $0.0500
Score Details
Visual Quality: 5.0/5 (100% conf)
The image demonstrates excellent visual quality, characterized by professional-grade sharpness, realistic textures, and effective lighting. There are no visible artifacts, distortions, or inconsistencies, making it appear highly natural and polished.
Input Fidelity: 5.0/5 (100% conf)
The transformed image exhibits excellent preservation of all key features, identity, and composition from the original input. There are no discernible differences in facial features, clothing details, lighting, or overall appearance, making the two images virtually identical.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved. The transformed image is an exact match to the original, with no discernible changes to facial features, skin tone, or expression.
face_similarity: 0.9894
firered-edit
Failed: Request 7dc7460c-f03f-4168-84ea-99c3c642a90f timed out after 300 seconds
flux-dev-i2i flux-dev-i2i
2.3/5
1824ms · $0.0250
Score Details
Visual Quality: 5.0/5 (90% conf)
The image exhibits excellent visual quality with a high degree of realism and no discernible artifacts. The subject's features, hair, and clothing are rendered with sharp detail and natural textures, contributing to a professional-grade appearance.
Input Fidelity: 1.0/5 (100% conf)
The transformed image completely fails to preserve any key features from the input. The subject's identity, gender, facial features, hair, clothing, and background are all entirely different, indicating a complete loss of input fidelity.
Identity Preservation: 1.0/5 (100% conf)
The transformed image features a completely different person with a different gender, facial features, hair, and overall appearance. The original individual is entirely unrecognizable in the output image.
face_similarity: 0.0000
⚠ Face similarity (0.0000) below threshold (0.6) — excluded from overall

Add dramatic studio lighting with a dark background

Input Input image
Reference image
reve-edit reve-edit
5.0/5
16892ms · $0.0400
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits excellent visual quality, with professional-grade sharpness, realistic textures, and sophisticated lighting. There are no visible artifacts, distortions, or inconsistencies typically associated with AI generation, making it indistinguishable from a high-quality photograph.
Input Fidelity: 5.0/5 (100% conf)
The transformed image perfectly preserves all key features from the original input. The subject's identity, facial features, clothing details, hair, and lighting are all maintained without any discernible alterations or losses.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved. All facial features, skin tone, hair, and expression remain identical between the original and transformed images.
face_similarity: 0.9752
seedream-edit seedream-edit
5.0/5
24831ms · $0.0300
Score Details
Visual Quality: 5.0/5 (95% conf)
The image exhibits excellent visual quality with sharp focus, realistic textures, and professional lighting. There are no visible artifacts, distortions, or blurring, making it appear as a high-quality photograph.
Input Fidelity: 5.0/5 (95% conf)
The output image perfectly preserves the subject's identity, facial features, clothing, and overall composition from the input. The only significant change is the dramatic lighting, which introduces a strong shadow across the face but does not alter or lose any underlying features.
Identity Preservation: 5.0/5 (100% conf)
The identity of the person is perfectly preserved. All facial features, hair, beard, and overall structure remain identical, with the only change being the lighting conditions which introduce a strong shadow.
face_similarity: 0.9084
flux-kontext flux-kontext
4.0/5
15895ms · $0.0500
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits professional photographic quality with excellent sharpness, realistic textures, and coherent rendering. There are no visible artifacts, distortions, or inconsistencies, making it indistinguishable from a high-quality photograph.
Input Fidelity: 3.0/5 (70% conf)
The transformation preserves the subject's identity, pose, and clothing, but significantly alters the lighting. This results in a much darker image where many facial details and the texture of the sweater are obscured by shadow, leading to a noticeable loss of visual information for these features.
Identity Preservation: 4.0/5 (80% conf)
The individual's core facial features, expression, and body proportions are excellently preserved, making the person instantly recognizable. However, the significant change in lighting between the two images leads to a noticeable difference in the visual consistency of the skin tone and complexion, appearing much darker and more shadowed in the transformed image.
face_similarity: 0.8455
firered-edit
Failed: Request ac7db757-a646-4931-a9ed-c6dc85b0f91a timed out after 300 seconds
flux-dev-i2i flux-dev-i2i
2.3/5
2101ms · $0.0250
Score Details
Visual Quality: 5.0/5 (100% conf)
The image exhibits excellent visual quality with sharp details, natural lighting, and realistic textures. There are no visible artifacts, blurring, or inconsistencies, making it appear professionally rendered.
Input Fidelity: 1.0/5 (100% conf)
The transformed image features a completely different person with distinct facial features, hair, and clothing compared to the original input. While the overall portrait style and black background are maintained, the core identity of the subject is not preserved.
Identity Preservation: 1.0/5 (100% conf)
The identity of the person has been completely changed in the transformed image. A different individual is depicted, making the original person unrecognizable.
face_similarity: 0.0755
⚠ Face similarity (0.0755) below threshold (0.6) — excluded from overall

Cost Summary

CategoryRequestsCost
fal.ai generation36$1.3050
gemini-2.5-flash judge108$0.0540
Local metrics$0.0000
Total$1.3590

Configuration

Modelsreve-edit, seedream-edit, flux-kontext, firered-edit, flux-dev-i2i
Judgegemini-2.5-flash
Dimensionsvisual_quality, input_fidelity, identity_preservation
Pipelineimg2img
Evalytic Version0.2.1
Platformdarwin-arm64
Metric Scoring clip_score: threshold=0.18, weight=0.2, lpips: threshold=0.4, weight=0.2, face_similarity: threshold=0.6, weight=0.2