ProGamerGov

@htoyryla > python scripts with neural-style, in order to do luminance only transfer Looking at the naming of the feature in Gatys' code, he calls it "Color Control", which leads...

@VaKonS Yes, the code converts the custom gray scale images back to the RGB after working with them in the LUV colorspace. I am currently finalizing my analysis of how...

> You seem to suggest that Gatys' code implements some other scheme than described in the paper. It can well be, but I am not convinced that your approach duplicates...

Comparing what I have learned/figured out, to what the [research paper says](https://arxiv.org/pdf/1611.07865.pdf), I think the linear-color-transfer.py script is for histogram matching: >If there is a substantial mismatch between the luminance...

Pulling images directly from Gatys' code (The iPython notebook example), the content images before the style transfer process are exactly the same as as the ones made by `--cp_mode lum`:...

@htoyryla > Like when you now say "after style transfer, you get pretty much the same output as with Neural-Style", I am confused because I believed your process used neural-style...

In my tens of thousands of tests, it's only happened to me a few times. Normally I see that the loss values went to extreme values before it happened, indicating...

@youyou0988 Neural-Style is a style transfer system, so you can't really avoid transferring the style. If you want to transfer the color from one image to another, I would suggest...

@htoyryla > I guess the name of the project is because these statistical color manipulation tools are meant to be used together with neural style transfer? Yep, the repository contains...

Doesn't Google Reverse Image Search already perform what you are trying to accomplish?