Qwen 2.1 - FineView Turbo (fixed) - INT8

Qwen 2.1 - FineView Turbo (fixed)

CHECKPOINT
Reprint


Updated:

Qwen 2.1 - FineView Turbo (fixed) by PrivateParty on Tensor.Art

From Civitai:

Built with Qwen* (an unofficial finetune of the Qwen-Image-2.1 VAE) from madebyollin

Texture-Fix-VAE-for-Qwen-Image-2.1 is the Qwen-Image-2.1 VAE, but finetuned to produce cleaner textures with no checkerboard artifacts (and no NaNs when running in fp16).

texture_fix_vae_for_qwen_image_2.1_bf16.safetensors

The curve function is correctly implemented, but ComfyUI made "calculating mu dynamically based on resolution" an optional feature instead of the default behavior, and they never implemented the official terminal stretching.

The former is a result of architectural layering (model configuration is fixed at load time, so you can't get the latent size at sampling time), and the latter is simply not done—they even gave Flux the same "static default + optional node" treatment.

If you want to get as close to the official standards as possible at 2K, the shortest path is: ModelSamplingAuraFlow(shift≈1.31) to compensate for mu, and you'll need ManualSigmas to eliminate any remaining terminal stretching differences.

Version Detail

Qwen-Image-2.1

Project Permissions

Model reprinted from : https://civitai.com/models/3003597

Reprinted models are for communication and learning purposes only, not for commercial use. Original authors can contact us to transfer the models through our Discord channel --- #claim-models.

Related Posts