#image-generation

Intermediate tutorial: Match your GPU's VRAM to the official H3-Image template or Pruna's P-H3-Image, update ComfyUI to 0.45.0 or newer, let your AI agent check disk space and place the four model files, then open the text-to-image and image-edit templates and start at 4MP. Includes 1 copy-ready prompt.

MiniMax released H3-Image, an open-weights image model that makes 4K pictures in about 10 seconds and accepts up to 16 reference images. Three ways to use it, what local hosting demands, and a prompt that lets your AI agent generate images for you.

ByteDance's Seedream 5.5 can edit the same image for 100 rounds while unchanged areas stay pixel-level identical. It makes more realistic photos, writes small Chinese and English text more accurately, and exports transparent PNGs. Includes two prompt templates.

Use when a user wants an agent to generate images, video, audio, 3D assets and game art, or to model, rig, build levels and create effects in Blender, Maya, ZBrush, Unreal, Unity or Godot. Built on the open-source scenario-labs/skills (MIT): run dry_run to estimate cost before generating, and confirm before spending; expert tools only need the matching software installed locally. Use case: Have an agent generate images, video, audio and 3D assets, estimate cost with a dry run before generating, and drive software such as Blender, Maya, ZBrush, Unreal and Unity.

OpenAI released Images 2.6 on November 4: from one sentence or one photo in ChatGPT you can generate animated GIFs, and export infinitely scalable, editable SVG vector files, about 50% faster. Useful for memes, icons, print materials and short video assets; SVG currently has limits: one image at a time, and text-only generation.