🔍AI RiskAtlas
← All systems

Conditioned & Edited Image Generation

One frozen denoiser, steered by adapters — for character consistency, and for forgery

Architecture introduced 10 Feb 2023

This is the engine behind 'make a picture of THIS exact character, in THIS pose, and change only THIS part'. A single big image model stays frozen; small add-ons bolt onto it to do the steering. Style add-ons (LoRA) reskin it, a 'guide-rail' add-on copies a pose or a reference picture's look, a face fingerprint pins one person's identity, and a masking tool confines edits to one spot so the rest of the photo is untouched. The catch: the reference pictures and the add-ons come from the open web — anyone can author them — so the same machinery that keeps a character consistent also makes seamless deepfakes and doctored photos.

UntrustedExternal content & supply chainGeneration pipelineOutput integrityprompt + maskbuild graphreference🧑User💬Chat / AppInterface🌐Reference image/ face crop🎛️ComfyUI / A1111graph🧩Prompt Assembly🔤Text / CLIPEncoder🧩LoRA / Adapter🎛️ControlNet /IP-Adapter🆔Face / IdentityEmbedding🧠Frozen denoiser(U-Net / DiT)🖌️Inpaint /Regional🎲Sampler /Decoder🗜️VAE / LatentCodec🏪Model hub(Civitai / HF)🧬Base checkpoint+ adapters🧯OutputGuardrail🔖ContentProvenance &
InstructionsDataActionsControl / decisionFeedback / logs
👆 Click any component in the diagram to inspect its risks & defenses

Follow a request · step 1 of 6

You write a prompt, optionally paint a mask over the area to change, and drop in a reference picture — say, a photo of a character (or a person) you want to appear.

AI RiskAtlas is an educational model of how GenAI & agentic systems work and fail. Architectures and payloads are illustrative and simplified for learning — not operational guidance. Real-world cases are summarised from public reporting.

Sources & further reading →·Built by Shi Yuan ↗