No model download active · Local GPU off
How to use (.txt)
Your images stay on this device
WEBGPU · NO INSTALL
Character modelling and animation

RiggerSet Studio

Generate a model, review repairs, add a rig and continue into animation.

Image

Load artwork and compare the prepared reference.

Steps change the view. Processing and downloads use their own buttons.

No artwork loadedNo model loaded
Graphics card
Choose a card available to this browser
Detect cards to choose one explicitly. Detection does not start generation.

Model files on this browser

These pretrained model weights are shared across characters. Each missing file needs its own download click. No model download starts when you enable processing.

Storage: this site → browser private files (OPFS) → ilgpu-ml-models. No Desktop folder, installer or ZIP is created. Another browser, profile or site address has separate storage.

Checking saved file metadata. No GPU or model download is used.

FLUX.2 Klein 4BOptional image / A-pose editing0 / 6 required files saved · About 4.6 GB

Source: LostBeard/FLUX.2-klein-4B-q4-ILGPU. Geometry files shown here match your selected 512 resolution. Existing alternate-resolution files are retained.

  • text_encoder_q4.safetensorsMissing from this browser’s model cache
  • transformer_q4.safetensorsMissing from this browser’s model cache
  • vae_decoder.onnxMissing from this browser’s model cache
  • vae_encoder.onnxMissing from this browser’s model cache
  • tokenizer.jsonMissing from this browser’s model cache
  • vae_norm.jsonMissing from this browser’s model cache
TRELLIS.2 + DINOv33D reconstruction, texture and regional edits0 / 14 required files saved · About 11 GB at one resolution; alternate resolution adds about 5.2 GB

Source: microsoft/TRELLIS.2-4B. Geometry files shown here match your selected 512 resolution. Existing alternate-resolution files are retained.

  • pipeline.jsonMissing from this browser’s model cache
  • ckpts/ss_flow_img_dit_1_3B_64_bf16.jsonMissing from this browser’s model cache
  • ckpts/ss_flow_img_dit_1_3B_64_bf16.safetensorsMissing from this browser’s model cache
  • ckpts/slat_flow_img2shape_dit_1_3B_512_bf16.jsonMissing from this browser’s model cache
  • ckpts/slat_flow_img2shape_dit_1_3B_512_bf16.safetensorsMissing from this browser’s model cache
  • ckpts/shape_dec_next_dc_f16c32_fp16.jsonMissing from this browser’s model cache
  • ckpts/shape_dec_next_dc_f16c32_fp16.safetensorsMissing from this browser’s model cache
  • ckpts/slat_flow_imgshape2tex_dit_1_3B_512_bf16.jsonMissing from this browser’s model cache
  • ckpts/slat_flow_imgshape2tex_dit_1_3B_512_bf16.safetensorsMissing from this browser’s model cache
  • ckpts/tex_dec_next_dc_f16c32_fp16.jsonMissing from this browser’s model cache
  • ckpts/tex_dec_next_dc_f16c32_fp16.safetensorsMissing from this browser’s model cache
  • ckpts/ss_dec_conv3d_16l8_fp16.jsonMissing from this browser’s model cache
  • ckpts/ss_dec_conv3d_16l8_fp16.safetensorsMissing from this browser’s model cache
  • model.safetensorsMissing from this browser’s model cache
SkinTokensSkeleton and skin weights0 / 1 required files saved · About 1.15 GB

Source: LostBeard/SkinTokens-bf16. Geometry files shown here match your selected 512 resolution. Existing alternate-resolution files are retained.

  • skintokens.safetensorsMissing from this browser’s model cache
MoveNet LightningSupport model for 17 pose landmarks0 / 1 required files saved · Additional small support model

Source: Xenova/movenet-singlepose-lightning. Geometry files shown here match your selected 512 resolution. Existing alternate-resolution files are retained.

  • onnx/model.onnxMissing from this browser’s model cache
All recognized cached files (0)

No recognized model files found in this site’s browser storage.

Removing these model files frees this site’s storage. Character projects, source images, exported models and unrelated cached models are kept.

STAGE 1 / 13

Reference

Start with your own full-body artwork. Original pixels stay in this browser.

pending

Asset type for your next upload

Choose the subject in your picture

The selected type is saved with your new project. Choosing a type does not enable the GPU or download files.

Sora production comparison

Load the exact AI-prepared front image originally used to make Sora. This preset skips another AI redraw, pose-model inference, repairs and repainting. It stops after generating the model for comparison.

Loading this reference does not enable GPU use or download model weights.

What to put in your image

  • One whole character, facing forward in a neutral pose.
  • Arms and legs separated; hands empty and clear of clothing.
  • Face, hands and feet fully visible, with an empty margin.
  • Transparent background or flat white; no scenery or cast shadow.
  • For hand-drawn art, use clear outlines and readable costume details.
  • One PNG, JPEG or WebP up to 10 MB and 24 million pixels. Only one input image is supported.

Your original file is kept as uploaded and used as the identity and outfit source.

Hidden anatomy and the back are inferred from this single view. Rebuilt hands, pose changes and held-item removal can be wrong and need your review. Preparation can attempt missing details; it cannot verify them.

Inference pixels are resized proportionally when needed to fit 4096 pixels per side. The shorter side must stay at least 16 pixels. Your saved original is unchanged.

Drop your character artwork here

One PNG, JPEG or WebP · up to 10 MB · full character in frame

Original files are immutable. Replacing artwork creates a new project.