Thanks a lot for the detailed explanation @mitsuki-youko! I think it's really helpful to see how others process work to try to see what works and what to adapt.
Haha, and yea @Gordon003, if it works, welp, it works. Don't touch it!
@Pattu_w yep, pretty much what I deal with on the inverse as well. I've tried different processes, but it seems there's no way around taking time to detail countours to some degree, either by filling in limits manually, erasing excess color spilling over, or taking time to create proper selection masks to fill in. Maybe it's something you get better with practice... I really wish there was a magic fill tool that understood what I want it to do.
@mitsuki-youko - This is all really helpful, and the video does show the advantages of using the lasso fill + masks + different layers to get those neat edgeless colors in place. I really appreciate the time taken to explain it in detail.
I should probably also explain a bit more of the pipeline I'm currently using, which should help identify how to improve it.
I have tried different things across the board, from using a single layer and paint everything on it (4 layers max, like Marc did in his zelda video), to breaking down every component to easily edit them later, which worked, but was almost as time consuming as cleaning up a single layer. The end result for both seems to be fairly good.
I've noticed they all come with advantages and disadvantages, but there's one pipeline I'm trying to refine, which is what I currently use.
1) Sketch. Not much to say, just a sketch, do whatever, then change it to a low opacity and set it behind to use as a base. Recently I'm taking notes from @patrycja.lerch and others and using a layer under the sketch to try out colors and overall composition before moving on.
I think the most important part here is to get the form and composition right, since colors and other details can be adjusted later, but form and composition will be harder to fix.
This is where I fixed that elbow. I also tendo to add some shading as per sketch practices to get an idea of how lighting will work. Gestures and what I've seen from others gesture exercices like value practice has helped improved my sketches here.
I might have different layers if I feel like it but I don't stress too much about it.
2) Basic inking. First pass. Lately with my lineart/inking exercices I've tried to make this ink pass as fine as possible. I'll usually bump up the resolution of my sketch, which I do at A5, up to 3 times, which brings the whole thing to A4. The first inking will usually still be a bit messy since I worry more about getting lines straight, and not worry about interesections. Once I'm done I'll come back and clean up the lines to get a clean ink setup.
Depending on what I'm doing, this might be my final ink pass, like for the manga, this is where inking ends (it's a more simplified process). For the manga since it's the only ink pass, i'll add some weight and details. Otherwise I'll keep the ink simple.
Only other point worth mentioning is that This is where I'll try to be as detailed as possible with each corner of the drawing. Something that stuck from the manga tutorials I've taken is that it's important to not leave botches of ink to add fuzziness (unless it's intentional, like in backgrounds), to try to be as clear as possible with each part. The nails, how muscles work, how clothes intersect and so on.
3) Flats. On their own layer of course. I've tried different things to do the flats, from using the lasso select/fill, lasso fill, manual painting, etc. What I found works for me at this stage is to add blotches of color that spill all over the place, and then "erase" the excess to get colors withing the lineart. It might seem like it would take a lot of time to clean up, but since the lineart is usually clean it's actually very easy and fast, for me it's faster than using lasso to fill in. I'll usually set my background to a high saturated color to easily notice missing gaps.
Once I'm done with flats I'll usually add a random background and play with hues a bit to see how they mix up.
4) Shading. I think this is where things start getting messed up. I will use flats and use the CSP option of Select color Gamut to select only that color, wich basically selects pieces of the piece to shade separatly. I'll also do laso selection and magic wand to select pieces as needed, then proceed to shade different parts together or independently.
Oh yea, I'm using a different layer for this.
I'm 90% certain this is where those white gaps are starting to be created. Perhpas I should expand the selection? or reduce the color gamut tolerance? Or just outright create masks manually?
Oh yea, I also "fill in" the flats on the back to create a white only layer that will later be used for the gradient maps.
At this point I might also try different shading techniques. For the manga it's a simple dark overlay to simulate shadows, for something like this piece, I tried to use a combination of round brush+blending+airbrush
5) Gradient Maps. I'll use the same selection tech from the previous step to create masks and apply gradient maps per color or per piece. Fairly straight forward, but I suspect these selection techs are what are messing me up, you can already see those white pixels here (though maybe not as clear in this low resolution post, but they are already noticeable in my canvas).
I found them!
these pixels will continue to get worse as more elements are added after this point.
6) Add details. Usually face details like the eyes, some more light in the hair with some glow, etc. I'll use a variety of layers as needed, but it's usually not more than 3 or 4, though I will usually adjust shading and lighting in a separate layer as well.
7) Some more details, like trying to do a proper background, rimlight and other details like overall saturation adjusment. This is when I'll usually notice the random pixels more and proceed to fix the whole thing. I also usually fix the lineart as needed, and depending on the piece I might lower opacity on it and multiply it, or draw over it with different tones to accentuate the lineart presence and make it thicker. It really depends.
I'm currently at this stage, and the only step pending would be to add further details as needed, like details on the clothes, adjust eye tones, etc. As @patrycja.lerch says, this is where I would let it "dry" for a few days or post it for feedback to see what else can be adjusted or compile lessons learned to keep in mind for the next piece.
By the end, there's more layers than I would prefer to have.
I think I'll to refine the whole process further. I'd be interested to see the process others do as well, and I keep watching videos in youtube of how others process work to see what can be learned from it.