Capitan01R/ComfyUI-Krea2T-Enhancer (new) "Virtual Image Conditioning"

v 1.5 Been experimenting with Krea2's Qwen3-VL vision encoder and stumbled upon a neat little trick. Basically, instead of feeding it an actual reference image, I'm feeding it a virtual canvas with a selectable RGB color. What's interesting is how much the color alone can influence the overall aesthetic of the generation. For example, purple works surprisingly well for nightclub scenes, while white or black tends to work better for portraits. Same prompt, different canvas color, noticeably different results. You can also control the virtual canvas resolution independently of the actual generat

<iframe src="https://silytics.si/embed/item/cf2a37b7-ed9b-4957-abcd-1eda874813fc" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Civitai and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.