TOBE2D Stable Cascade TEXT to IMAGES V1 — v1.0

TOBE2D Stable Cascade TEXT to IMAGES V1 — v1.0 is a Stable Cascade text-to-image workflow: enter a text prompt, use Stage B and Stage C models, and create image output that can be saved.

The workflow covers prompt entry, text encoding, sampling, VAE processing, image output, and saving through CR Prompt Text, CLIPTextEncode, KSampler, VAEEncode, VAEDecode, CR Image Output, and SaveImage.

Enter a text prompt in CR Prompt Text; CLIPTextEncode is the named text-encoding component.

The listed Stage B and Stage C model files are stable_cascade_stage_b.safetensors and stable_cascade_stage_c.safetensors.

CR Image Output is the named image-output component, and SaveImage is the named image-saving component.

Update ComfyUI to the latest version before using the workflow.

Suggestion · non vérifié

Choose one Stage B and Stage C model pair for your VRAM: stage_b + stage_c, stage_b_bf16 + stage_c_bf16, stage_b_lite + stage_c_lite, or stage_b_lite_bf16 + stage_c_lite_bf16.

Suggestion · non vérifié

Named node packs include ComfyUI_Comfyroll_CustomNodes, ComfyUI-mape-Helpers, and cg-use-everywhere.

Listed model files also include dreamshaperXL_v2TurboDpmppSDE.safetensors.

Other named nodes include CheckpointLoaderSimple, CLIPTextEncode, FreeU, KSampler, SaveImage, Seed Everywhere, StableCascade_EmptyLatentImage, StableCascade_StageB_Conditioning, VAEDecode, and VAEEncode.

The description lists 5 steps for Stage C and 3 for Stage B; it also names 10 steps in Stage C and 5 in Stage B as the best combination found.

The workflow includes an added Face fixer.

The creator describes this as a personal experimental release for speed and quality.

No source is listed for dreamshaperXL_v2TurboDpmppSDE.safetensors; verify where to obtain it before setup.

Suggestion · non vérifié