Add Anima-2.9B (40-block expanded) checkpoint

Add official anima-base-v1.0 as default checkpoint

Serve output as lossless PNG instead of webp

Fix prompt weighting examples (no bracket de-emphasis in ComfyUI core)

Drop backend info line from info block

Add info block below image: backend, loaded models, live GPU/VRAM stats, weighting note

Add er_sde/dpmpp_2m_sde_gpu samplers, er_sde default, README-aligned negative prompt

Default ZeroGPU duration to 30s (like mrfakename/anima-v1)

Bound w/h, steps and CFG like mrfakename/anima-v1 for stable generation time

Add gradio progress bar via ComfyUI sampling callback

Drop invented model labels, use real filenames

Simplify: drop LLM enhancer, examples and verbose text

Anima Collection: Gradio Space over ComfyUI backend (oai/civitai-collections models)