In the newly published book it states that the color spaces of the source material should be linear ones and matching. It also states that if the content is in Rec.709 (or sRGB, which is the same i guess?) it should be kept that way.
Yet from what i searched up Rec.709/sRGB isn't a linear color space, so should there be made extra changes to the input material and if so, what would be a good way to do it? In Da Vinci Resolve I tried to change it, but it would end up in a total mess that looks really wierd.
So currently I extracted material from youtube video sources, one side is 1080p resolution, the other side 2160p including .jpg image sources from the web.
Both have the same Rec709 color space as far as the checkup-tool (from https://mediaarea.net/de/MediaInfo) says, yet the higher resolution side takes longer to train and is less detailed in the previews. Would it make sense to scale down the 2160p to 1080p? or is the problem color space related?