Combined workflow (ComfyUI txt2img, Wildcards, LLMs/Ollama, Pony, SDXL, Illustrious, Flux, Qwen, Z-Image, Anima, Flux 2 Klein, Krea 2 Turbo) — Work-in-Progress

This ComfyUI workflow turns text prompts into images with Illustrious, Pony, or SDXL, then refines and upscales each result to 16MP; optional Wildcards and an LLM/Ollama extension can expand prompts.

Stage 1 generates an image and sends it to a selector, which can be bypassed for batch generation.

Stage 2 uses HiResFix to upscale to 4MP. Stage 3 uses Ultimate SD Upscaler (No Upscale) with the original model and LoRAs, then applies face and eye detailers.

Stage 4 sends the image to SeedVR2 to produce the final 16MP result.

Wildcards processing and an LLM/Ollama prompt extension are optional; Ollama can also create a multiline narrative.

The workflow produces CivitAI-compatible metadata for each image-generation stage.

Consult the included READ ME FIRST section for usage details.

Suggestion · non vérifié

Node packs listed include cg-image-filter, comfy-image-saver, ComfyLiterals, ComfyUI_ADV_CLIP_emb, comfyui_ultimatesdupscale, comfyui-custom-scripts, and comfyui-easy-use.

Also listed are comfyui-fbcnn, comfyui-image-saver, comfyui-impact-pack, comfyui-impact-subpack, comfyui-logicutils, comfyui-ollama, and rgthree-comfy.

Named model files include bbox/Eyeful_v2-Paired.pt, bbox/face_yolov8m.pt, ema_vae_fp16.safetensors, Illustrious/illustrijGEN_1.safetensors, remacri_original.safetensors, sam_vit_b_01ec64.pth, SDXL/sdxl_vae.safetensors, and seedvr2_ema_3b_fp16.safetensors.

For batch generation, bypass the selector.

Suggestion · non vérifié

To unload the Ollama model after a request, set “keep alive” to 0.

Suggestion · non vérifié

The author says they are stopping maintenance of the full combined workflow in favor of smaller workflows.

Ollama use may be constrained by available VRAM.