XaiJu
scarygoround
scarygoround

patreon


Q&A 2: Remakes

Aram writes:

"I see artists remake their work quite frequently. I remember seeing some redrawn panels by you (maybe for the printed version of [Giant Days]?) Have you considered recreating some of the vector looking comics from the mid noughts? Why/why not?"

I wanted to write a little bit about the drive to remake old work. It's an instinct that, carefully managed, can service improvement, or it can be a very destructive process that signals creative problems that can't be fixed by doing variations on the same old thing.

When I first started making webcomics around mid-1998, one of the few people putting stuff online was a guy called Steve Troop, who had a comic strip called Melonpool. Steve's work was professional-looking (at a time when almost no one in the field was) and charming. But I can remember (I think) two separate occasions where Steve went back to the start, hundreds of comics into his run, and started redrawing the strips from scratch. Even the later iteration of the comic was a reboot that repurposed old situations. I believed then and am certain now that this was a bad way to go. You can't be so in love with your little world, yet so sure that it's wrong, that you endlessly reprocess it looking for some different result. I don't want to put Steve down - he's a great cartoonist. His evident technical skill made it stranger to me that he couldn't find what he was looking for, but went back for it anyway. But I took note. I'd rather put something out that's 95% finished than spend my life noodling over the last 5% and never putting it out. 

BUT I do redraw comics sometimes, but for a couple of different reasons:

1. I want to try out a new art technique.
Old comics are a good source of material to try out new things. I can try out inking techniques, pencilling techniques, bugged out layouts, weird colour ideas, format changes or anything else without having to write new material. If I manage to finish the page, I tend to share it (nowadays, usually here) because it's a bit of bonus fun. 

As an added extra, the original page provides a basic rough to work from.

2. I have a shortage of new material.
A few times, I've repurposed old stories in a slightly glossier format. One was the Heavy Metal Hearts and Flowers book I did for Keenspot around 2004, which built out a fairly slight Bobbins series from a few years earlier into a book-length story. I made those comics while simultaneously making Scary Go Round daily strips, and doing freelance design work, so I had to repurpose something old. There are a couple of other examples of this - the Girlspy minicomic (from 2003) and Oldbourne, a story I redrew while writing Giant Days and By Night for Boom in 2017.

So in answer to Aram's question, while I've redrawn one or two vector strips (referring to the old slick-looking Scary Go Rounds from 2002-2006) for my own amusement, I find them hard to work from as roughs (the compositions were limited), so I tend not to pick them out when I'm testing something new. 

It's fun to mark your progress by looking at then-and-now versions of things, but it's counterproductive to look down on past work and always assume that what you're doing now is 100% better. My remakes are often missing something. Innocence sometimes looks better than crusty old experience. 

--

Next time, I'm going to write about writer's block, and the mixture of household chemicals that will cure it -- every time.

If you have a question you'd like to see answered, about art or writing or anything else creative, please ask - in the comments, via email, or a Patreon message. 


Comments

I'm not a follower of the Star War.

This is brilliant. Also appreciate the lack of questioning who shot first.

The AI stuff is very new. It is great stuff for repeatedly raising your hopes to great heights, and then dropping them to see if you break. But it is beginning to do useful jobs in the motion picture industry. Give it a few years, maybe.

I find this all very interesting, don't get me wrong! I use every tool I can to help me. But I do go through spells of wishing I'd never started using any of them because when I don't... you can tell.

That's when they try to get AI to do the whole thing for a fancy paper. Computers are good at doing the boring, grunt jobs. Take your earlier example of colouring up to a black line. Someone writes a program to do flood fill up to the nearest black edge. With care, they can program to cover all the special cases - what to do with grey levels, and holes within the black lines. Or you can give the program some examples of before (holes) and after (flooding done the way you want it) and the thing can then do it the way you want. AI is great for rotoscoping and stuff like that. identifying people, and in all rotoscoping, for example. We use it for tracking objects and people, and drawing outlines around them. This is a terrible job to do for 50 frames a second, but AI can get most of the outlines right first go, and where it doesn't you can poke it in the right direction. There was a set of cards that let you randomly write a tune in the style of Mozart. Back in the day Mozart himself had a set and loved it. Adding a texture in the style of someone else is nothing new. I feel new tools could speed up your cleaning-up passes, so you can do more art and less pixel-bashing.

Back in the day, when cab drivers would still speak to you, I would sometimes be drawn on my job. Among the three responses that eventually led me to lie and say I was in plumbing supply sales by default, was the supposition that "it's all done with computers these days, isn't it?" While no doubt, one day (doubtless in my lifetime) it will all be done with computers, AI is fortunately still not at the point where machine learning can reverse engineer my decisions, any more than those songs where AI has been trained on the entire Frank Sinatra catalogue, sound like anything anyone would listen to. Personally, I think the desire on the part of computer scientists to remove human endeavour from art - the inevitable market-driven result of the process - borders on sociopathic.

Artificial Intelligence can do a thing called Art Style Transfer. See for example... https://towardsdatascience.com/art-style-transfer-using-neural-networks-a28f5888746b. ...or even ... https://medium.com/analytics-vidhya/generating-japanese-style-animation-using-deep-learning-not-cyclegans-or-neural-style-transfer-1b48faeba931 This works best with still images that are similar in style. It should be possible to produce Bobbins Vol 1 in the style of Steeple. More intriguingly, if you have the rough sketches and the finished work, it might be able to do the same. Whether the results are good enough to use is anyone's guess but it might be fun to try. It's a big project but someone might take it on. I am working with AI but with motion pictures: this is rather different and not the sort of thing I could do in my spare time just yet.


More Creators