Miguel’s likeness & proportions

MiguelCompare Miguel’s likeness with his references. A realistic face alone is not a match.

“Dream bigger” · dialogue

Same generated starting image. Kling delivered; Seedance rejected the reference.

Kling 4.0 Omni

10.042s · 1080p · 24 fps

References used

Reference details

Saved V4 voice used in Kling; exact waveform preservation is not claimed. Original photographs and source voice stay private.

Est. $8.79/minEstimated · $1.4652 for this test

Cost basis

135 actual Kling credits. USD estimate uses the displayed $32.56 / 3,000-credit plan; not an invoice. Rate normalized to requested 10s; delivered 10.041667s. Fees/taxes unknown.

Miguel’s note: “A lot of it was random words.” Requested speech: “Dream bigger.” Exact transcript has not been verified.

AI score70/1005/5 reviewed

Judge-by-judge reviews

1 Yes · 2 No · 2 Unsure. Range 60–90/100. AI diagnostics; community votes and Miguel’s verdict stay separate.

ClaudeNo60/100
anthropic/claude-sonnet-5.5
Limited assessment: brief explanation shortened; full text retained. Identity, setting and Saturn are well preserved, but the performance departs from the brief: continuous mouth movement across most of the clip instead of a single short phrase after a pause. A watermark is also visible. Not usable as is.
GeminiNo70/100
google/gemini-4-argon
The visual likeness, lighting, and environment match the reference well, with natural hand and head motion. However, the subject appears to speak across three separate intervals rather than delivering one brief phrase and resting as requested.
GrokUnsure61.25/100
x-ai/grok-4.7
The output preserves the seated man, spacecraft cabin, and distant Saturn with generally strong visual continuity. However, the requested hand gesture appears only briefly and disappears unnaturally, while sampled frames cannot confirm the specified pause, exact speech, or voice.
QwenYes90/100
qwen/qwen3-vl-235b-a22b-instruct
Face, body, visual quality, and temporal continuity fully match reference. Shot accuracy and physics meet target with minor unverifiable deviations (audio/lip sync excluded per brief). All visible elements preserved without defects. Output suitable AS IS for visual-only use per brief.
MistralUnsure82.5/100
mistralai/mistral-large-4-0
Strong visual match to reference image with preserved face, glasses, hair, beard, navy crewneck, and spacecraft cabin setting with distant Saturn. Static camera and waist-up framing correct. However, critical brief deviation: speech appears continuous throughout rather than 'brief pause then says exactly once' — mouth is active in most sampled frames from ~1.3s to ~9.3s, suggesting either multiple

Video frame reviews cover sampled visuals; they do not certify audio, lip sync or every frame.

CommunityVotes are recorded separately below

Voting help

If your browser blocks embedded sign-in, open this review separately ↗. Votes and Credits use the same Navo account.

Prompt, settings & test details

One submission; no retry. 135 Kling credits were charged. The USD conversion is a plan-based estimate, not verified cash billing.

Seedance 2.5 through fal rejected the same generated image under its real-person likeness policy (HTTP 422). No video, no retry; recorded API charge $0.00.

Human likeness and lip sync acceptance are not formally recorded. No new AI judges were run for this layout update.

Rejected attempt · Seedance 2.5

Submitted, rejected, no output. $0.00 API usage; no visual score or ballot.

Full-body images

4 images · 4 models

Same photos and prompt. Compare each result with its references.

Sort

How it sorts

Within each test: highest complete AI rating, then lower historical cost. Partial panels stay unranked by rating. Lowest cost reverses that priority. Unknown values follow known ones. Ranges use their upper estimate. Human votes stay separate.

Video uses USD per delivered minute. Images use USD per generation until complete spend and human usability checks establish cost per usable image. Failed submissions and retries stay in the test history. Fees and taxes are unconfirmed.

Seedream 5.0 Pro ↗ · 2

References used

Side profile

Reference details

Original photos; backgrounds edited and arranged for display. Test inputs unchanged.

Original photo; background edited for display.

Est. $0.1485/imageEstimated · $0.1485 for this image

Cost basis

Estimated usage $0.1485; actual billing unconfirmed. Native 1536 × 2048.

AI scoreReview neededRecords found

Previous AI reviews

0 of 5 judges returned usable numeric scores. The combined score waits for five. Unknown results are not zero. AI diagnostics; community votes and Miguel’s verdict stay separate.

Archived source records found. Prompt and reference associations need review before a corrected judging packet is qualified. The earlier panel used an incomplete packet (5/5 reviews received). Fresh reviews have not run yet.

ClaudeUnsure—
anthropic/claude-sonnet-5.5
The image is a clean, realistic full-body studio shot that largely follows the brief's wardrobe, pose and background. Face and body fidelity are essential but cannot be verified without references, so use as-is is undecided.
GeminiUnsure—
google/gemini-4-argon
The image shows a coherent, realistic studio portrait matching the general wardrobe and pose, though the oversized T-shirt visually lengthens the torso against instructions. Because the four required reference photographs are absent, face and body fidelity cannot be verified.
GrokUnsure—
x-ai/grok-4.7
The photograph visibly satisfies the requested pose, wardrobe, framing, and studio presentation, with clean realistic rendering. However, the original identity and body references are unavailable, so the essential face and proportion requirements cannot be confirmed.
QwenYes—
qwen/qwen3-vl-235b-a22b-instruct
Limited assessment: face rating excluded: required reference or brief unavailable. Still-image time offsets ignored. body rating excluded: required reference or brief unavailable. Face likeness, body proportions, shot accuracy, visual quality and physical coherence all meet or nearly meet targets. Minor body deviation (shirt hem) doesn't break fidelity. Output is suitable AS IS for the brief.
MistralUnsure—
mistralai/mistral-large-4-0
The generated image successfully executes the technical and compositional requirements of the brief: correct pose, wardrobe, lighting, background, camera perspective, and physical coherence. However, the two most critical dimensions—face likeness and body proportions—cannot be verified because the reference photographs (Reference1 for body, References2-4 for face) were not supplied. Without these,

Video frame reviews cover sampled visuals; they do not certify audio, lip sync or every frame.

Earlier reviews · 1

Gemini 3.8 Flash · 75/100
Earlier visual rubric · 2026-10-13. Earlier model/rubric; excluded from this five-judge score.

CommunityVotes are recorded separately below

Voting help

If your browser blocks embedded sign-in, open this review separately ↗. Votes and Credits use the same Navo account.

Prompt, settings & test details

Same four original references and shared prompt. One candidate; no retries.

Likeness and proportions awaiting owner review. No winner selected.

bytedance/seedream/v5/pro/edit

Estimated usage $0.1485; actual billing unconfirmed. Native 1536 × 2048.

Google Nano Banana Pro ↗ · 1

References used

Side profile

Reference details

Original photos; backgrounds edited and arranged for display. Test inputs unchanged.

Original photo; background edited for display.

Est. $0.16/imageEstimated · $0.16 for this image

Cost basis

Estimated usage $0.1600; actual billing unconfirmed. Native 1792 × 2400.

AI scoreReview neededRecords found

Previous AI reviews

0 of 5 judges returned usable numeric scores. The combined score waits for five. Unknown results are not zero. AI diagnostics; community votes and Miguel’s verdict stay separate.

ClaudeUnsure—
anthropic/claude-sonnet-5.5
The image follows the shot brief well and is clean and realistic. Without the reference photos, face likeness and body proportion fidelity, both essential, cannot be verified, so the decision is unsure.
GeminiUnsure—
google/gemini-4-argon
The image satisfies the studio setting, wardrobe, and standing pose requirements with realistic quality and coherent anatomy. However, because the four reference photographs are missing, facial likeness and exact body proportion fidelity cannot be verified.
GrokUnsure—
x-ai/grok-4.7
The photograph closely matches the requested studio composition and appears visually coherent, but the required historical references are absent. Facial identity and body-proportion fidelity therefore cannot be verified, preventing approval as-is.
QwenYes—
qwen/qwen3-vl-235b-a22b-instruct
Limited assessment: face rating excluded: required reference or brief unavailable. body rating excluded: required reference or brief unavailable. Output fully satisfies the brief: face likeness preserved with exact features and asymmetry; body proportions match reference1’s compact torso and longer legs; shot framing, wardrobe, lighting and perspective are accurate; visual rendering is realistic without artifacts; anatomy and shadows are physically coherent. No major or minor deviations detected.
MistralUnsure—
mistralai/mistral-large-4-0
The generated image follows the technical shot requirements well: correct wardrobe (black crewneck tee, black shorts, barefoot), proper stance (straight-front, relaxed, arms at sides, hands visible), white seamless background, centered framing, waist-height camera perspective, and coherent anatomy with natural proportions. However, without the four original reference photographs, I cannot verify:

Video frame reviews cover sampled visuals; they do not certify audio, lip sync or every frame.

Earlier reviews · 1

Gemini 3.8 Flash · 75/100
Earlier visual rubric · 2026-10-13. Earlier model/rubric; excluded from this five-judge score.

CommunityVotes are recorded separately below

Voting help

If your browser blocks embedded sign-in, open this review separately ↗. Votes and Credits use the same Navo account.

Prompt, settings & test details

Same four original references and shared prompt. One candidate; no retries.

Likeness and proportions awaiting owner review. No winner selected.

gemini-3-pro-image

Estimated usage $0.1600; actual billing unconfirmed. Native 1792 × 2400.

GPT Image 2.5 · Sunburst ↗ · 3

References used

Side profile

Reference details

Original photos; backgrounds edited and arranged for display. Test inputs unchanged.

Original photo; background edited for display.

Est. $0.40/imageEstimated · $0.4 for this image

Cost basis

Estimated usage $0.4000; actual billing unconfirmed. Native 1536 × 2048.

AI scoreReview neededRecords found

Previous AI reviews

0 of 5 judges returned usable numeric scores. The combined score waits for five. Unknown results are not zero. AI diagnostics; community votes and Miguel’s verdict stay separate.

ClaudeUnsure—
anthropic/claude-sonnet-5.5
Image is a clean, realistic full-body shot that follows the wardrobe, pose and background of the brief. Without the reference photos, face and body fidelity, which are essential, cannot be confirmed.
GeminiUnsure—
google/gemini-4-argon
The image satisfies the studio setup, pose, wardrobe, and realism requirements. However, the four reference photographs are missing from the input, making it impossible to verify face likeness and exact body proportions.
GrokUnsure—
x-ai/grok-4.7
The still matches the requested wardrobe, stance, framing and white studio setup and looks visually coherent. Without the original reference photographs, face likeness and body-proportion fidelity cannot be judged, so it should not be accepted as-is.
QwenYes—
qwen/qwen3-vl-235b-a22b-instruct
Limited assessment: face rating excluded: required reference or brief unavailable. body rating excluded: required reference or brief unavailable. Face likeness and body proportions are preserved per references. Shot accuracy, visual quality, and physical coherence fully meet brief. Minor body proportion deviation is acceptable. Output is suitable AS IS.
MistralUnsure—
mistralai/mistral-large-4-0
The generated image follows the brief's technical specifications well: correct wardrobe (black tee, black shorts, barefoot), proper stance (straight-front, arms relaxed, hands visible, knees straight), white seamless background, centered framing, waist-height camera perspective, and coherent anatomy with fingertips at upper-mid thigh. However, without the four original reference photographs, I无法独立

Video frame reviews cover sampled visuals; they do not certify audio, lip sync or every frame.

Earlier reviews · 1

Gemini 3.8 Flash · 75/100
Earlier visual rubric · 2026-10-13. Earlier model/rubric; excluded from this five-judge score.

CommunityVotes are recorded separately below

Voting help

If your browser blocks embedded sign-in, open this review separately ↗. Votes and Credits use the same Navo account.

Prompt, settings & test details

Same four original references and shared prompt. One candidate; no retries.

Likeness and proportions awaiting owner review. No winner selected.

gpt-image-2.5-sunburst-2026-09-08

Estimated usage $0.4000; actual billing unconfirmed. Native 1536 × 2048.

FLUX.2 Max ↗ · 4

References used

Side profile

Reference details

Original photos; backgrounds edited and arranged for display. Test inputs unchanged.

Original photo; background edited for display.

Est. $0.88/imageEstimated · $0.88 for this image

Cost basis

Estimated usage $0.8800; actual billing unconfirmed. Native 1536 × 2048.

AI scoreReview neededRecords found

Previous AI reviews

0 of 5 judges returned usable numeric scores. The combined score waits for five. Unknown results are not zero. AI diagnostics; community votes and Miguel’s verdict stay separate.

ClaudeUnsure—
anthropic/claude-sonnet-5.5
The image matches the shot brief well and is photographically convincing, but face and body fidelity cannot be verified without the reference photos. Using it as is remains undecided.
GeminiUnsure—
google/gemini-4-argon
The output follows the general studio setup, wardrobe, and standing pose, though the low shirt hem visually elongates the torso against the brief. Because the four required reference photographs are absent, face likeness and body proportion accuracy cannot be verified.
GrokUnsure—
x-ai/grok-4.7
The image closely matches the specified studio shot and appears visually coherent, but missing references prevent verification of the essential face likeness and body proportions. It cannot be approved as-is on the available evidence.
QwenYes—
qwen/qwen3-vl-235b-a22b-instruct
Limited assessment: face rating excluded: required reference or brief unavailable. body rating excluded: required reference or brief unavailable. Output fully satisfies brief: face and body match reference anatomy, pose and setting are accurate, visual quality is high and coherent. No major defects in any dimension; all targets met with minor deviations absent.
MistralUnsure—
mistralai/mistral-large-4-0
Limited assessment: Still-image time offsets ignored. The generated image matches the technical shot requirements well: correct wardrobe (black crewneck tee, black shorts, barefoot), proper stance (straight-front, relaxed, arms at sides, hands visible at upper-mid thigh), white seamless background, centered framing with appropriate margins, soft studio lighting, and coherent anatomy. However, critical evaluation of face likeness and body proportions—

Video frame reviews cover sampled visuals; they do not certify audio, lip sync or every frame.

Earlier reviews · 1

Gemini 3.8 Flash · 75/100
Earlier visual rubric · 2026-10-13. Earlier model/rubric; excluded from this five-judge score.

CommunityVotes are recorded separately below

Voting help

If your browser blocks embedded sign-in, open this review separately ↗. Votes and Credits use the same Navo account.

Prompt, settings & test details

Same four original references and shared prompt. One candidate; no retries.

Likeness and proportions awaiting owner review. No winner selected.

fal-ai/flux-2-max/edit

Estimated usage $0.8800; actual billing unconfirmed. Native 1536 × 2048.

About this test

Four images returned. Judge face likeness and body proportions; usable rate and final costs are unconfirmed.

Check face shape, head scale, build and leg length together. Native settings differ; a visually preferred attempt is not a universal model winner.

Population performance, confirmed usable rate and settled costs remain unconfirmed.

Character sheets

3 sheets · 4 submissions · October 5, 2026

Four body views and four face studies from the same five photos and prompt. One submission failed.

Sort

How it sorts

Candidate C · Google Nano Banana Pro ↗

2752 × 1536

References used

Est. $0.1471/imageEstimated · $0.1471 for this image

Cost basis

Generation estimate USD0.147062 before taxes; actual billed charge unconfirmed.

AI scoreReview neededRecords found

Previous AI reviews

0 of 5 judges returned usable numeric scores. The combined score waits for five. Unknown results are not zero. AI diagnostics; community votes and Miguel’s verdict stay separate.

Archived source records found. Prompt and reference associations need review before a corrected judging packet is qualified. The earlier panel used an incomplete packet (0/5 reviews received). Fresh reviews have not run yet.

ClaudeNot reviewed—
anthropic/claude-sonnet-5.5
GeminiNot reviewed—
google/gemini-4-argon
GrokNot reviewed—
x-ai/grok-4.7
QwenNot reviewed—
qwen/qwen3-vl-235b-a22b-instruct
MistralNot reviewed—
mistralai/mistral-large-4-0

Video frame reviews cover sampled visuals; they do not certify audio, lip sync or every frame.

Earlier reviews · 1

Gemini 3.8 Flash · 75/100
Earlier visual rubric · 2026-10-13. Earlier model/rubric; excluded from this five-judge score.

CommunityVotes are recorded separately below

Voting help

If your browser blocks embedded sign-in, open this review separately ↗. Votes and Credits use the same Navo account.

Prompt, settings & test details

gemini-3-pro-image · Direct paid Google

Native 2K · 16:9 · JPEG returned

27.03s

Generation estimate USD0.147062 before taxes; actual billed charge unconfirmed.

Candidate A · Seedream 5.0 Pro ↗

2560 × 1440

References used

Est. $0.153/imageEstimated · $0.153 for this image

Cost basis

Generation estimate USD0.153000 before taxes; actual billed charge unconfirmed.

AI scoreReview neededRecords found

Previous AI reviews

0 of 5 judges returned usable numeric scores. The combined score waits for five. Unknown results are not zero. AI diagnostics; community votes and Miguel’s verdict stay separate.

ClaudeUnsure—
anthropic/claude-sonnet-5.5
Well-executed realistic eight-panel sheet that follows layout and wardrobe. Without references, identity and body fidelity cannot be verified. Face crops cut the crown and show the shirt collar, which deviates from the brief.
GeminiUnsure—
google/gemini-4-argon
The image successfully generates a clean, photorealistic eight-panel turnaround sheet matching the requested views and layout. However, the shorts are noticeably shorter than the requested knee-length, and because the source reference images are missing from the input, facial and anatomical likeness cannot be verified.
GrokYes—
x-ai/grok-4.7
The sheet fulfills the requested eight-panel reconstruction with consistent appearance, correct views, and clean photographic presentation. Missing references prevent likeness scoring, but no visible defect blocks use as is.
QwenYes—
qwen/qwen3-vl-235b-a22b-instruct
Limited assessment: face rating excluded: required reference or brief unavailable. body rating excluded: required reference or brief unavailable. Output fully complies with brief: all eight panels show the same person with accurate face and body proportions, correct wardrobe, and studio lighting. Layout, framing, and visual fidelity match requirements. No visible defects or deviations.
MistralYes—
mistralai/mistral-large-4-0
Limited assessment: face rating excluded: required reference or brief unavailable. Still-image time offsets ignored. body rating excluded: required reference or brief unavailable. The output is a strong multi-view character reference sheet matching the brief's layout, wardrobe, and view specifications. All eight panels show the same adult male with consistent facial features, body proportions, black t-shirt and charcoal shorts, bare feet, and pure white background. The 65/35 row split, four-column arrangement, and specified view angles are correctly implemented. Face close

Video frame reviews cover sampled visuals; they do not certify audio, lip sync or every frame.

Earlier reviews · 1

Gemini 3.8 Flash · 75/100
Earlier visual rubric · 2026-10-13. Earlier model/rubric; excluded from this five-judge score.

CommunityVotes are recorded separately below

Voting help

If your browser blocks embedded sign-in, open this review separately ↗. Votes and Credits use the same Navo account.

Prompt, settings & test details

bytedance/seedream/v5/pro/edit · fal.ai

Standard prompt enhancement · 2560×1440 PNG

119.323s

Generation estimate USD0.153000 before taxes; actual billed charge unconfirmed.

Candidate B · GPT Image 2.5 · Sunburst ↗

2560 × 1440

References used

Est. $0.2721/imageEstimated · $0.2721 for this image

Cost basis

Generation estimate USD0.272136 before taxes; actual billed charge unconfirmed.

AI scoreReview neededRecords found

Previous AI reviews

0 of 5 judges returned usable numeric scores. The combined score waits for five. Unknown results are not zero. AI diagnostics; community votes and Miguel’s verdict stay separate.

Archived source records found. Prompt and reference associations need review before a corrected judging packet is qualified. The earlier panel used an incomplete packet (1/5 reviews received). Fresh reviews have not run yet.

ClaudeUnsure—
anthropic/claude-sonnet-5.5
A realistic eight-panel sheet that mostly follows the layout and wardrobe. Identity and body fidelity cannot be verified without references. Shorts color and face crops deviate from the brief.
GeminiNot reviewed—
google/gemini-4-argon
GrokNot reviewed—
x-ai/grok-4.7
QwenNot reviewed—
qwen/qwen3-vl-235b-a22b-instruct
MistralNot reviewed—
mistralai/mistral-large-4-0

Video frame reviews cover sampled visuals; they do not certify audio, lip sync or every frame.

Earlier reviews · 1

Gemini 3.8 Flash · 75/100
Earlier visual rubric · 2026-10-13. Earlier model/rubric; excluded from this five-judge score.

CommunityVotes are recorded separately below

Voting help

If your browser blocks embedded sign-in, open this review separately ↗. Votes and Credits use the same Navo account.

Prompt, settings & test details

gpt-image-2.5-sunburst-2026-09-08 · Direct OpenAI

Max quality · 2560×1440 PNG

68.353s

Generation estimate USD0.272136 before taxes; actual billed charge unconfirmed.

Candidate D · FLUX.2 Max ↗

No output

Route rejected the input size.

No image; no retry.

Reference not released for this test.

Cost/image unknownCharge unverified

Cost basis

Rejected job: actual billed charge/refund unknown; no image returned.

Prompt, settings & test details

fal-ai/flux-2-max/edit · fal.ai

Standard prompt enhancement · 2560×1440 PNG

6.903s

Rejected job: actual billed charge/refund unknown; no image returned.

Requested area too large. The total allowed size for input and output is 10 megapixels. The total input size is 7 megapixels after downsizing the first image to a maximum of 4 megapixels, and subsequent images to a maximum of 1 megapixels. With an output size of 4 megapixels, the requested total is 11 megapixels. Please adjust your input images or output size and try again.

Original photos

Original framing and proportions. The display board is cropped and downscaled.

Full body ↗Front ↗Profile ↗Smile ↗Rear hair ↗

Method, costs & limits

Each model received the same five original photos and prompt for four body views and four face studies. White cutouts are display edits, not generation inputs.

Native 2K-class outputs vary in size. No retries, repairs or upscaling. Seedream enhances prompts internally. Rear-body anatomy and speaking motion remain untested.

All four submissions count. Human acceptance, final charges and refunds remain unknown. Successful-generation estimates total $0.572 before taxes. Voting is unblinded; community and AI scores stay separate.

How to compare these tests

7 delivered images from 8 submissions, including 1 failed submission. Each test used its own reference pack and output format. Compare models within a test; scores and votes stay attached to the original attempts.

This groups existing results. No images were regenerated, no final likeness was approved and no combined project score is calculated.

Try this shot

Plan a full-body reference check for your own character. Approve costs before rendering.

Miguel’s likeness & proportions

“Dream bigger” · dialogue

References used

References used

References used

References used

References used

References used

References used

References used

Try this shot

AGV Report