These are LoRAs that tested whether color development could be controlled by conceptual manipulation. It doesn't use any image information, it tries to make the colors more vivid just by concept (whether it actually changes or not may depend on the model).
●Vivid-I ver.2.0
I have added a new version of Vivid-I. The previous version focused on increasing color, but this one focuses on improving contrast. Also, the previous version did not emphasize colors close to skin tones, but this version increases saturation without distinction.
-------------------------
●Vivid-C
This is an experimental LoRA that brightens colors using only conceptual information.
It affects the composition a little. Affects the character's clothing, etc. This can be used when, after repeated merging, you end up wearing only monochrome clothes.
●Vivid-S
LoRA makes pictures colorful with conceptual information. It consists of many more concepts than Vivid-C. Although it only includes concepts related to color, it is quite relevant to the composition. Vivid-C
By using a large value, it can also be used as a secondary effect LoRA. The upper limit is about 3.
Here is an example of how it was used as a special effect. I honestly don't really understand what's going on. This is a special effect that would be useless for normal characters, but it may be useful for some characters.
*@Vivid-S:3.0
●Vivid-I
This is LoRA, which aims to sharpen colors based on images rather than concepts. The conceptual version affected the composition more than I expected, so I also created an image version.
By making it smaller than the conceptual version (only 916KB), I aim to maximize the effect while minimizing the overall impact. This is a very ordinary slider, but in practical terms this may be the most useful.
I compared it with the LoRAsI have.
Saturation Slider
描述:
训练词语:
名称: Vivid-C.safetensors
大小 (KB): 6705
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
