Log in
Andromeda Chapter 6 debriefing 2
Hey crew, getting this out late, planned to get it out the day before Halloween, but I ended up not having the time to do that. Starting this off with by saying this one will be a bit different then other briefings because it covers both September and October due to be not getting one out on time for September.
Starting off as usual with script work. Script work across both months was about par for the course most months, but could have been better, totaling 300 lines of code, 123 in September and 177 in October. This would put us at averaging 4.10 lines per day for September and 5.71 for October. And a month over month change of 4% and 44% percent respectively for total lines and for average lines per day a month over month change of 8% and 39% respectively.
For render work, September wasn't really up there, because that wasn't the focus, script work was generally more important as a focus at the time so only 14 renders in all for September, including the extras for the National Cheeseburger Day 2023 extra, resulting in a -30% month/month change.
For October, I ended up deciding to try something a bit different. Usually I setup a file that covers a single location in a single scene, so sometimes these files are 3-4 frames while other times they maybe 10-14 frames. It just depends on how many different expressions and poses I need for that location at that scene. Also sometimes I want different camera angles for the same shot so those bring the count up further beyond the number of frames. Only after those are all set do I set the file to render each frame. I decided to instead focus on drafting each of the renders, creating those files with the frames that comprise the images that actually get rendered. So outside of the actual renders for the Halloween 2023 extras for Captain tier supporters I didn't render anything in October. Instead what I can do is this month and December, I'll likely release the update near the end of December, I can load those files and render them each in turn. For the frames that have been made and need to be render, I haven't really kept exact track of them, but its around 40 at this time.
On a related note to renders, and I might have hinted at this in a previous post but I'm to lazy to go back and check. As I have mentioned in the past I do the renders using my personal computer, the same one I use for playing games, watching YouTube, etc. as a result rendering can interfere with my own use of my computer, unsurprisingly this mostly impacts gaming. Several months back I bought myself some older server hardware to use for a Plex server, which I have completed, but an idea came to mind that I should be able to slap a workstation GPU, think something like the Nvidia RTX A4000, in there and use it for rendering Daz3d Iray.
At a basic level, the most important specs for Daz3d are VRAM and Cuda cores. VRAM impacts how large and detailed your scene can be and cuda cores impact how quickly iray can render, clock speed also matters to an extent but most tests I have seen on the daz forums seem to indicate that number of cuda cores matters much more.
To compare the specs of the A4000 to my RTX 3080 12GB. The A4000 is:
Starting off as usual with script work. Script work across both months was about par for the course most months, but could have been better, totaling 300 lines of code, 123 in September and 177 in October. This would put us at averaging 4.10 lines per day for September and 5.71 for October. And a month over month change of 4% and 44% percent respectively for total lines and for average lines per day a month over month change of 8% and 39% respectively.
For render work, September wasn't really up there, because that wasn't the focus, script work was generally more important as a focus at the time so only 14 renders in all for September, including the extras for the National Cheeseburger Day 2023 extra, resulting in a -30% month/month change.
For October, I ended up deciding to try something a bit different. Usually I setup a file that covers a single location in a single scene, so sometimes these files are 3-4 frames while other times they maybe 10-14 frames. It just depends on how many different expressions and poses I need for that location at that scene. Also sometimes I want different camera angles for the same shot so those bring the count up further beyond the number of frames. Only after those are all set do I set the file to render each frame. I decided to instead focus on drafting each of the renders, creating those files with the frames that comprise the images that actually get rendered. So outside of the actual renders for the Halloween 2023 extras for Captain tier supporters I didn't render anything in October. Instead what I can do is this month and December, I'll likely release the update near the end of December, I can load those files and render them each in turn. For the frames that have been made and need to be render, I haven't really kept exact track of them, but its around 40 at this time.
On a related note to renders, and I might have hinted at this in a previous post but I'm to lazy to go back and check. As I have mentioned in the past I do the renders using my personal computer, the same one I use for playing games, watching YouTube, etc. as a result rendering can interfere with my own use of my computer, unsurprisingly this mostly impacts gaming. Several months back I bought myself some older server hardware to use for a Plex server, which I have completed, but an idea came to mind that I should be able to slap a workstation GPU, think something like the Nvidia RTX A4000, in there and use it for rendering Daz3d Iray.
At a basic level, the most important specs for Daz3d are VRAM and Cuda cores. VRAM impacts how large and detailed your scene can be and cuda cores impact how quickly iray can render, clock speed also matters to an extent but most tests I have seen on the daz forums seem to indicate that number of cuda cores matters much more.
To compare the specs of the A4000 to my RTX 3080 12GB. The A4000 is:
- A single slot card and has 16GB GDDR6 VRAM (448GB/s Memory Bandwidth)
- 6144 cuda cores
- Total Board Power 140W
While my RTX 3080 12GB is:
- A 2.75 (just say 3) slot card and has 12GB GDDR6X VRAM (912.10GB/s)
- 8960 cuda cores
- Total Design Power (so not apples to apples but I can't find TBP for my card and its "close enough") of 350W
So the A4000 has a bit more VRAM, making larger scenes easier, but less cuda cores, and is roughly (in theory) half the power draw. Because they are also much smaller I could in theory slap two inside a server chassis and have more cuda cores then my personal rig with about the same amount of power draw. Only draw back is, the A4000 is kinda expensive, looking at anywhere between 800 and 1000 USD, granted I spent about 800 on my 3080 but its still expensive. So this idea is something that probably won't be done anytime within the next couple of months, but maybe down the road after some combination of more support and/or either a much better paying job or a really good raise.
Now someone had mentioned in the Discord server (https://discord.gg/KCp6u4ekEh), which isn't super active but I do try to respond when people say or ask stuff:
...chapters could be little longer but this is not exactly surprising as u work alone.
That is something I agree with, and chapters have gotten a bit longer with more recent updates due to me getting in the habit of saying "Screw it, I won't be able to get all I want done in a single release unless I have it be a year between updates."* I hopeful that this 3-4 month release schedule I plan to use going forward (which is something I've said in the past I wanted to do, but I've not been able to stick to it before) will enable me to have longer updates since I won't have as much of an issue with asking myself "Do I want to end this chapter here or try for another few hundred lines for the next good stopping off point?"
*I personally don't like waiting a year for an update, but I'm not looking to change how others due updates so I'm not going to say anything else on that note beyond "Do what works best for you."
On that, that is going to be it for this debriefing, this was probably the longest one I have done to date, most of it though being that idea to potentially help with renders in the future. I will see you all near the end of this month, or early next month, with another debriefing.
