May Update

Sorry for the lack of communication lately. Here’s the monthly update for May (that’s only 15 days late…).

What to expect for this month
I am planning to complete Digital Infection (tier 3). If all goes well, I’ll be moving onto the next comic which is Fusion Hungry Ch.2 (tier 1). The comic order priority is listed on my Trello Board. Each tier will be rotated so if I worked on tier 1 first, the next month tier 1 will be worked on last and tier 2 and tier 3 comics will be worked on next.

Comic Backlog
I have 16 comics I’m behind on. I’m still playing catchup, but I’m happy to almost be done with all of October comics. Currently have 4 October comics to complete, then I can move onto the comics planned for this year. (Added the delayed comics to October month). Progress is moving slower than I want, but it’s still progress moving forward.

Workflow Update
I’m changing my workflow. Previously I worked on multiple comics at one time. I’m switching now to focus and finish one comic chapter before moving on to the next. 

I’ve finished all the comics created on Comic Life 3. Making the full switch to Affinity Publisher to do my comics and I’m loving how much more control I have over the comic and effects. One thing that took awhile was redoing uncompleted comics from Comic Life 3 to Affinity Publisher. Now all future comics will be done with Affinity Publisher so the time spent on that should be faster.

Messaging
The google doc links will be messaged at the end of the month, that way I don’t have to resend messages. May google doc links haven’t been sent yet, probably around the last day of the month or the first day of the next month.

You can skip reading the segment below. I'm just explaining a lot of the things behind the scenes and technical jargon.

Good News
I learned how to optimize my scenes. I have reduced texture files before, but I didn’t like how it kept saving into my Daz library and I didn’t quite understand how to fully utilize the scene optimizer script. Did some learning while I did Project Skins Ch.1 and the results came out good. I did several test renders. 

I can fit 6 figures into the scene. These were independent figures, but I’m not sure if the textures get reused or anything. These were not instances since instances mimic the parent object. Instances are basically a copy of the parent object, so whatever I do with the parent object, the instances will change to mimic the parent object’s new pose or material or anything.

Closeup render test.

Failed render (scene exceeded 2080ti 11gb VRAM)

I did another test trying to fit the scene with 8 characters. This didn’t fit into the 2080ti so the result was no render. (The above picture, which is just a black image.)

Successful render image with reduced texture maps.

I did the same test that failed, but this time I reduced all the figure’s maps by half. Basically optimizing the scene to fit into the graphics card. And the scene was able to render.

All images were rendered in 4K resolution. I’m not sure how big of an impact it has on eating the VRAM, but I was surprised that it was able to fit this much characters even at 4K.

I’ll need to do more renders and see how it fits into the scene. Only the figures' texture maps were reduced to 2K since they all come mainly in 4K. I’ve read somewhere that each surface of the figure references from the texture maps. So if there is one texture map that has multiple surfaces, then it gets loaded again and again into the graphics card, resulting in more VRAM usage.

I have tried this before and the results weren’t too good for me. I was clueless on how to save the texture files so I can load it back in if needed. But that was almost a year ago and now I can use this under my belt.

I have been wanting to upgrade my PC for sometime, but right now isn’t the best time. The next thing I could do is learn how to optimize my scene to fit more characters. There are several things I needed to keep in mind. I prefer to keep my scenes lightweight if possible so it’ll be faster to load and save.

The only drawback I have with more characters is I have to manage more characters. But the good thing is I now have the ability to go for a larger scene with more characters if I wanted to. I always wondered how games can do it and realized that they often optimize their games to be able to render at many frames. However, it is different with Daz3D since the Iray render engine focuses on rendering an image as accurately as possible. There is a difference between a game engine and a render engine. I’m not sure about the specifics. This is just what I’ve read online.

I also used another product along with the scene optimizer. I used a shader resource saver which basically replaces all the Genesis figure maps with a basic shader. 

In the club scene in Project Skins Ch.1, I used the shader saver resource particularly on the bartender, singer, and extra guy. These are easy solution to fit in more characters into the scene without eating too much VRAM. There is a loss of detail with these shaders though.

PC Hardware Research
I want another 2080ti since more cuda cores would equal faster renders. A Titan RTX would be preferable for the 24gb VRAM, but that price is really not affordable for me anytime soon. A quadro card such as the RTX 8000 would be amazing for the 48gb, but yeah… I could only dream of having four Nvidia quadro RTX 8000... I would need to win the lotto to get that dream build. 

I’m eager to see the next generation card that Nvidia will offer. Hopefully Ampere doesn't cost as ridiculous as Turing, but I ain’t getting my hopes up yet...

Did more research on my build too and realized that my motherboard only supports 2 graphics cards and 64gb RAM. I currently have 32gb RAM and I’ve been noticing that Daz3D eats the system ram to max once I hit render. I’ll likely go with 128gb RAM if possible. 256gb RAM is want I originally wanted, but I choked looking at the prices of RAM... Even 128gb RAM makes my heart ache... I’ve been looking into a future build consisting of a motherboard that supports 4 graphics cards to speed up render time. I’ve seen what one 2080ti with +4000 cuda cores can do. I just can’t imagine what the render time would be with +12000 cuda cores. Course the VRAM would be a limiting factor, but since I learned how to optimize the texture maps properly, I think it’s a big game changer allowing me to do more with what I have. 

I’ve researched this a lot and decided that it’s best to go with something I can afford while not losing an arm and a leg… I'll have to wait a little longer since right now isn't the time for an upgrade.

Oh yeah, I recently cleaned my computer too! It was so dusty. And Jeez… My cable management was horrible. I changed my case to optimize more airflow since there was a module I wasn’t using. Had this computer since 2018 and the dust buildup was so ugly to look at, yet very satisfying to clean.

I hope everyone is doing well!

-AstralBot3D 2020.05.15