Apologies if all the technical stuff sounds too complicated, but don't worry because that's all hidden from the end users. Basically, animators could just draw 2 key frames to make their animations. That sounds fairly user friendly to me. If you want lots of animation freedom, and 2 key frames is too limiting and not worth the effort, try to envision this as the seed of something bigger in the future.
GMan003:This is a component of "animated avatars are cool".
Glad you agree! Everything is a component of something else, even the opposing arguments.
That is a concession and not an argument.
Doh! I thought about that ex post facto last night. What I meant was "Keeping both animators and non-animators happy through different forum modes." That's a benefit that outweighs the loss of dissatisfying one side or the other.
You've identified a subset of acceptable avatars, not an exhaustive set.
I agree with this. The point of the subset is to start minimally small and make moderation easier, and naturally that makes animation easier.
Just doing proper color delta is some fairly hefty math to be doing on every pixel of every frame, even just during upload.
Can we precalculate how many CPU cycles it would take to compare two frames? Avatars are 15,625 pixels. So we double that for 2 frames, 31,250. Then multiply that by each operation in the equation, 8. I get 250,000 operations. Then we might roughly double that again for the tweening, so 500,000 operations. At this rate, even on a low-end 2GHz server, 4000 animations could be processed every second. (correct me please, that sounds overly optimistic)
Update: I see my error now. Need to factor in IPS not GHz.
Xenocartographer:My gut feeling is that actual algorithmic solution would be equivalent to a general-purpose image recognition algorithm, and if you can solve that, you have bigger things to do than argue on a comic site, like revolutionizing major industries and making billions of dollars.
Yes! Google is already the leader in this field, and they provide an API to their Vision A.I.
The pricing is fairly cheap too. We could perhaps train a neural net to do this, but we'd need a ton of example GIFs. High tech like this often trickles down into common usage eventually.
I've really enjoyed this discussion no matter what the outcome. It's absolutely true that animation is complicated and challenging on every level. Most people don't bother with it. But like JFK said, we do these these things "not because they are easy, but because they are hard."
That's a good point. User friendliness is important, and I think that clear guidelines on the upload page would help in that regard. In order to help animators, maybe show 4 squares on the upload page, representing frame 1 and frame 3 as the key frames, and 2 and 4 as the tween frames.
Clear instructions always seem helpful and user friendly. Maybe some beginner animation training tutorial that explains what tweening is. And maybe highlight examples that circle how 10 percent of pixels have changed between frame 1 and 3.
Maybe it would be more friendly to upload frames one at a time, rather than all at once. If they don't pass, write to browser with the reason why. If approved, then stitch them together automatically.
In exceptional cases, I suppose users could request personal moderation to override the automated process.
GMan003:Z3) Even if an objective rule can be written, the technical work to enforce it via code would be substantial, contributing to (B2)
Those are well-reasoned arguments that conclude the costs outweigh the benefits. Based on the limited benefits listed in A1-A2, it's impossible to disagree with that. On the other hand, other benefits may be gained:
1. New and valuable technical knowledge
2. New strata of Comic Fury functionality
3. New ways of community engagement and satisfaction
4. Fostering a community of animators that focus on quality through subtlety
5. Attracting new users that enjoy animated avatars
6. Keeping animation-opposed users happy through the non-animated mode
In terms of technical work, some open source stuff has already done the heavy lifting. I suppose the work would be accomplished slowly at some point in the distant future, as Kyo has other priorities and interests right now.
In terms of objective rules, we've made some headway into that as well. By identifying the sine wave cycle as the ideal model for subtlety, we can approximate that with 4 sample frames. Moreover, this contributes to the simplification of automated moderation, because frames 2 and 4 should be identical.
Another criterion is the difference in change of pixels between frame 1 and frame 3, which should be no more than 10 to 20 percent. Honestly, I don't know how to do that yet. But I'm not sure that the time/effort cost required to learn that technique is so prohibitive that it's not worth the benefit of learning and implementing something new.
I did a little research on that question and found that Manga Studio 5 was actually a rebranded version of Clip Studio by Celsys. Keep in mind that the EX version has more features than the PRO version.
That's part of why I voted YES (with caveats), because it's possible to have a non-animated forum reading mode (the default we have now), and alternatively an animated forum reading mode. The two parties never have to see each others' mode if they don't want to. I like the idea of an extra dimension of space-time layered on top of the old reliable static Comic Fury.
Hopefully Matt Comics won't mind, but I want to analyze the avatar he kindly shared, in terms of possible animated avatar rules.
I uploaded it to ezgif.com/split, and confirmed it has only 2 frames. One problem with only 2 frames is that it's a binary condition, which means the motion is flashing on and off, like a square wave. Now, for this avatar it's fairly subtle because the brightness of the glow frame is relatively close to the blue sky.
However, for helping Kyo set standardized rules that can cover a wide range of animations, I propose a minimum (and probably a maximum) of 4 frames. This way, you have a minimum sample to approximate a sine wave, which feels more analog, natural and smooth. What's cool about this is that any animator can start with drawing only 2 frames (a resting state, and a peak state), then 'tween blend between them to create the intermediary frames. This tweening only has to be done once, then copy it on either side of the peak state frame. ezgif.com's editor has a 'crossfade frames' effect which accomplishes this easily.
And voila! A total of 4 frames, animation smooth, subtle.
I think this one is good because it loops in logical continuous motion, kind of like you'd expect a real person to move. So this might be another criterion to consider. No choppy resets.
Here's another one homegrown on Comic Fury. This is for The Metallic by JuicyGrey. A beautiful animation for his web comic. Maybe a little too fast for a forum avatar, but really nice work. You can almost feel that breeze. I love how the shadow shimmers subtly on the ground too.
Great point. This is a vote for "subtly animated avatars," not anything goes ones, so we need examples and well-defined criteria. If everyone doesn't mind sharing some subtly animated avatars that you like, that would help voters get a visual aid on what they're voting for.
For consistent frame and duration criteria, consider this:
60 seconds static initial frame<br />
1 second animation of 5 frames<br />
Loop back to first frame
(consistent colors would probably mean no color changes allowed. so no strobing)
For pure ideal consistency, we would have to have 100% identical users, so all the avatars and content would be the same.
The content of avatars varies widely already, but they're limited by size when we upload them. If it's too big, it's automatically rejected. Therefore, avatars do have consistent boundaries that are enforceable automatically, by the server script.
My contention is that the animated boundaries can be enforced automatically as well. It's possible to automatically limit frame quantity and the time duration of frames. It's probably possible to limit color changes too.
Maintaining the current consistency would be easy enough by making non-animated avatars the default forum reading mode.
The animated mode need never be seen at all by those opposed. They won't even see a difference.
(Currently, two different forum reading modes are already in operation, AFAIK.)
As an aside, the problem with pausing GIF animation in situ is that every entire multiframe GIF gets downloaded regardless of user preference. So that costs server resources and bandwidth for the client too.
If the single frame is scraped just on the upload page, that script won't be running all the time during normal forum usage.
In agreement with you, Text, so that's why it's okay to have it both ways.
I guess in terms of environmental ethics, animated avatars are objectively BAD, because they would burn somewhat more electricity to process. Having them turned off by default would mitigate that, and having them as an option would satisfy the other half of the community. (the Animated-Avatar-Friendly Coalition, the AAFC)
If server resources are objectively overloaded by (scraping the first frame from) GIFs, then it's more clearly an absolutely BAD option to have them. In that case, there's no reason to consult the community, so should be a top-down decision.
It's a marvelous drawing but the coloring seems too dark to my eye. Maybe you can print it out to see how it looks in CMYK. I wonder if it needs highlights like a rim light on the right side. How important is this character in your comic?
My inclination is to make the banner identical to the cover of the book, as a marketing tactic, because you want people to recognize the book if they see it again somewhere, and that familiarity is how you build a brand. I'm thinking iconic like Captain America, the Grinch, or Harry Potter, for examples, so you want to make your main character, cover and logo iconic.
You've got a great name for a comic artist, so definitely put your humongous signature on the banner as well!
(Like Stan Lee, Dr. Seuss, J.K. Rowling)