Welcome to TVPaint Community forums !

I can give some tips from my experience that may save time/backtracking/frustration:
-Use your final cleaned up animation.
-Make sure beforehand that all gaps are closed. You can also just add another animation layer to add lines to close gaps, add those to your CTG source. It's annoying to have to switch back and forth if you missed gaps and takes time because your palette needs to be reopened if you want to change colours.
-I usually make the first frame with all colours so that I have the palette complete.
-Make sure "apply changes" is unchecked, otherwise you have to wait for tvpaint to calculate every time you touch an area.
-Then I use the exclusion colour first to go through all frames and fill the spots I don't want filled. Otherwise it's harder to see.
-I repeat this with the next colour and go through all the instances to fill in that one colour. I might fill multiple parts with the same colour, usually I concentrate on areas that belong together or just one part.
-A line across the surface area is usually enough, you can also run it through multiple areas.
-After all the colours are set, check "apply changes" and grab a drink.
-Check to correct any errors.

I've also found it useful to turn off the lightning bolt / 'apply changes' icon in the layer and then go and add CTG strokes for all the frames, and then click 'apply changes' and wait for the CTG brain to do its thing, and then turn it off again and do another pass, apply changes, etc. until it's done. It takes too long to wait for CTG to think for every stroke. Once you get a sense of how the CTG system processes fills, you can kind of prep a lot of it before applying the changes.

I'm curious if CTG happens on the processor, or graphics card? Because I'm thinking of upgrading my computer hardware, mostly to be able to colour faster as it takes me ages to get through colouring, and is generally un-fun slog-work.

If there ever were a use for AI in animation, this would be it - the computer understanding the relative relevance of areas between frames and the user just needing to colour in the first frame of the scene and the computer figuring out the rest. A kid would be able to identify parts of a body on a character, eg 'head', 'hair', 'shirt' so a computer should be able to do the same.

mikdog wrote: ↑21 Jul 2024, 12:17 I've also found it useful to turn off the lightning bolt / 'apply changes' icon in the layer and then go and add CTG strokes for all the frames, and then click 'apply changes' and wait for the CTG brain to do its thing, and then turn it off again and do another pass, apply changes, etc. until it's done. It takes too long to wait for CTG to think for every stroke. Once you get a sense of how the CTG system processes fills, you can kind of prep a lot of it before applying the changes.

I'm curious if CTG happens on the processor, or graphics card? Because I'm thinking of upgrading my computer hardware, mostly to be able to colour faster as it takes me ages to get through colouring, and is generally un-fun slog-work.

If there ever were a use for AI in animation, this would be it - the computer understanding the relative relevance of areas between frames and the user just needing to colour in the first frame of the scene and the computer figuring out the rest. A kid would be able to identify parts of a body on a character, eg 'head', 'hair', 'shirt' so a computer should be able to do the same.

I'm quite a big fan of the old method of CTG filling, via the old TVP10 plugin "Lazybrush" back then you always had to press a big button to initiate the algorithm, the only issue is that it never smartly calculated areas not meant to be filled so you always had to include the exclusion colors for it to operate.

Thanks. Mmm, I still need to do a fair bit of ‘exclusion’ marking with the current CTG layers, unlike a normal fill, ctg doesn’t seem to understand ‘holes’ in fills particularly well and seems to often need exclusion indicators. I start from biggest to smallest filling, or ‘outside to inside’ filling generally as that seems to be the best way to work with ctg layers.