Detect photorealistic and stylized images generated by Tencent Hunyuan (HunyuanDiT). Analyze skin micro-texture, background geometry, specular highlights, and synthetic diffusion residual noise.
Visual Forensics Guide
Hunyuan produces strikingly photorealistic and stylized images, but its diffusion-transformer decoder still leaves the characteristic fingerprints of a diffusion model:
Hunyuan faces look airbrushed — skin is rendered smooth and even, with pores and fine detail that are too regular and lack the asymmetric blemishes of real photography.
Straight lines in architecture, furniture, and tiling bend or drift near the subject, and repeating patterns melt together on close inspection.
Highlights on skin, lips, and surfaces read as glossy and plastic, sitting on top of the image rather than integrating with its material and light.
Object boundaries and flat regions carry faint halos and high-frequency residual noise typical of latent diffusion-transformer decoders.
Targeted forensic checklists, generator signature scanners, and catfish verification guides.
Instant screenshot paste (Ctrl+V) & AI profile detection.
Expose synthetic personas & stolen model profile pictures.
Identify OBS virtual camera & DeepFaceLive injections.
Is your online match real? Verify video proof & greetings.
Detect cloned voice notes & distress audio scams.
Protect against bot syndicates & fake ID verification.
Expose pre-recorded circular video note loops & bots.
Detect unauthorized face replacement & body doubles.
Inspect facial boundary seams & skin tone discrepancies.
Inspect videos for DeepFaceLive & FaceFusion synthetic replacement.
Spot live deepfake calls on FaceTime, WhatsApp & Zoom.
Universal detection across Sora, Kling, Runway, Luma & deepfakes.
Spot Sora 2 synthetic physics & temporal morphing.
Detect Ray 1 & Ray 2 camera drift & reflection flaws.
Analyze 1080p generation patterns & kinetic interaction.
Identify synthetic diffusion curves, camera velocity & AI video clips.
Detect DeepMind Veo 2 cinematic camera motion & diffusion artifacts.
Detect Pika Labs cinematic motion & Pikaffects artifacts.
Detect WanX 14B/1.3B 3D VAE artifacts & prompt-diffusion melting.
Identify Tencent dual-stream diffusion patterns & camera drift.
Spot THUDM 3D causal VAE edge ringing & temporal anomalies.
Inspect micro-facial muscle jitter & hair feathering.
Expose lip-sync latency & celebrity face swaps.
Spot virtual AI models & ambient loop inconsistencies.
Scan viral storytelling clips for AI diffusion artifacts.
Detect AI generated songs, vocal clones & audio diffusion.
Spot micro-skin pore uniformity & synthetic catchlights.
Spot GPT-4o image diffusion, plastic skin & style quirks.
Analyze SynthID traces and Google Imagen diffusion artifacts.
Verify open-weight diffusion checkpoints & anatomical glitches.
Spot Grok 2 candid flash textures and synthetic noise layers.
Identify hyper-cinematic rim lighting & teeth geometry.
Spot cloned voices, synthetic timbre & pitch cadence anomalies online.
Catch ElevenLabs voice clones & synthetic audio narration.
Analyze vocal tract resonance & missing respiratory pauses.
Verify photos sent in WhatsApp chats & reverse lookup.
Learn the hand-wave test to expose live call deepfakes.
Expose VoIP burner lines, Twilio numbers & scam networks.
Browse biometric-verified creators — zero bots or AI fakes.
Tired of fake personas and AI bots?
On Sealed Rose, every creator is verified through biometric facial matching and government ID. Zero AI models. Zero catfish.
