For dithering it uses some simple color quantization so the actual max amount is 216. And if you change the amount you have to re-select the option (as in select something else and then back to the option because I forgo to set the event), for it to show up.

Deltarune

At least I got my edit exercise done. Didn't do any sound effects or fany graphics but the focus was on cut & a bit of fusion to get more used to Resolve. Used some #sts2 footage and I think it came out not too bad.

That liveticker is a goldmine

Let's try a snapshot for better resolution

I do like this effect, although it is hard to control the details and eats memory with how inefficient I made it in Resolve.

Also I love these Birbs, even though i maybe could have picked an image that better shows the potential problems with median cut.

This took longer than I thought it would, mainly because I kept going back to my code and reading articles about color stuff to not write anything wrong. Gotta make some images and then there will be a post about dithering, and soon™ even dithering in my gif editor (just some UI stuff needed).

POV: You are abstracting while Caine asks about your opinion on the last adventure.

I really dig this effect, a bit like a torn poster. This happens when I do not do something like grayscale the intial image via luma or soemthing like that. The direction comes from diffusion matrix used

last but not least, my k-means version vs a bayer dither using median cut for pallete creation. Now I gotta make a bit of UI so that everything is nice and selectable in my gif editor.

So, the problem with my aproach to error diffusion is that some colors can get completly overblown because the distances are not equal/the palette is not balanced. Here is Pomni, using a random color pallete, once with 8x8 bayer dither and once with Floyd-Steinberg.

At least the Bayer dither works first try 🥳 No error diffusion should mean I do not need to take that much care about the colors being balanced, so in theory I can do something really dumb whioch will not work with FS or other aproaches.

Bonus BW Caine

got it; the main problem was that the color tables were not balanced, leading to some areas dipping so far into one direction that they simply would not come back. I think skipping the "classic" version (with a simple threshold) at first was what led to my initial confusion. Now I can sleep.

At least I have a generalized function I can jsut slot in whatever matrix and colorpallete I want. gonna do a dumber version of this tomorrow which should work.

Still some problewms with dithering, although i think I know where I am doing things wrong. The results so far are at least visualy interesting when I use the median cut color table.

Now that is better. Just got a < and > switche dup so my median cut was selecting empty sets to cut after a while. Now I can play around with what metrics to use. At the moment it's just plain amount of colors, but something like highest distance to the average color could be cool too.

Played around with more palette reduction stuff because my dithering will need to rest before I try it again. This is 16 colors for some reason, but should be 256 colors; still looks kinda neat.

POV: your code still does not work but at least it looks kinda cool.

Better but not quite there yet. I had an idea what I did wrong before, but now I am looking at my code with a big question mark over my head.

Tried to implement basic dithering, and I think I did something wrong.

More playing around with Resolve's fusion tab. There are some things I want to add like a wobble for free floating elements and maybe some smal splashes/displacmeent at the bottom. So far I find the animation part a bit convulted, like it needs extra steps when compared to AE.

Hier sowas. Die Dinger sind sauch eher so ein Notnagel, weil der Unterdruck natürlich warme luft wiede reinzieht. Es gitb auch varianten mit einem zweiten Schlauch wo dann Außenluft genutzt wird für den Abtransport der Wärme; die sind auch effizienter.

Goodbye Adobe, it was okay knowing you.

At least I got something fun done tonight while trying to get a bit more experience in fusion (ignore the animtion error on the hair)

Pomni. Look, I know we don't get along, but please put the gun down.

what better than spending a few minutes before going to bed and using the test from earlier today to play around some more. The text interaction did take a moment to figure out and I think could be done better than my current solution (each splat exists a second time and uses the text as a mask)

I like using expressions when testing stuff in AE, but it causes things to sometimes be quite inefficient. Take this example of some color splatter.