Well look at that, I managed to edit together a video and post it to YouTube. Honestly speaking in terms of video quality, if I were trying to be a content creator generally, this video would probably be a black mark against any quality standards I might seek to establish, but that’s not what’s going on here; the publishing of this video is actually an accomplishment in and of itself that marks the end of a long series of struggles. Let’s go over the journey thus far.
The First Steps
My setup is a bit different from the average person for a few reasons. The biggest complicating factor here is that my primary storage is actually a network share and not local to my personal machine. The original video editing software I was using wouldn’t connect to a network share without shenanigans, let alone handle the video in a way that was conducive to making proper edits, and it also didn’t seem to handle how large video files have gotten very well, so starting off this journey, my tools were dull even in optimal conditions and any editing would be an extreme compromise of something; an equation I didn’t want to attempt balancing. So I just didn’t; I decided I’d revisit the idea of editing videos when I had the time and patience to look at new software.
Eventually when I came back around to editing, I landed on DaVinci Resolve mostly because it was free and I didn’t (and still don’t) know exactly how far I’ll go with editing. If I end up figuring out a workflow and getting over my reservations related to how long editing takes, I’ll likely pay for a license, but the fact that it allows me to trial basically all the functionality I could ever need to make a proper edit for free is a massive score setter really. As an added bonus, DaVinci Resolve was able to see video files on a network share, and I was actually able to edit together a clip from a stream:
You should check out the clip, mostly because I think Fionna is funny and charming, but one thing you may notice is that the available resolution only goes up to 720p. I am one of those people that gets driven mad if I’m not able to work with footage in what I would consider good quality, and as such I hate working in resolutions below 1080p, however the source stream this clip is from is 720p because Fionna wasn’t streaming on her usual setup, so I made the clip in 720p instead of trying to fool around with upscaling or anything and it worked out quite nicely. This experience would deceive me into thinking my setup would work properly in every situation.
So brilliant, right? I can edit now, everything is amazing and I can get right onto my next project…right? The next project I wanted to work on involved cutting down a series of 6 hour streams. No problem, I’ll just load the video in the same way I handled this last one and…after playing back for a few seconds, the video desyncs and playback is so choppy that I can’t make accurate cuts. And changing the preview window size has no effect on this behavior. And moving the source footage to my local drive has no effect on this behavior. And attempting to generate proxy media immediately causes my primary storage device to run out of space.
Problems, problems, problems. What happened here is likely that the video file from YouTube that I worked with on the original clip was in a format that DaVinci Resolve plays nicely with, however the other streams I wanted to cut down were from Twitch and probably encoded differently with one or more of those differences causing this disjointed behavior.
The eventual solution to all of this was actually to change the configuration of DaVinci Resolve so that way it would generate proxy media on my network share instead of my local computer. If you’re not familiar, “proxy media” is what DaVinci Resolve calls files that has been re-encoded in a way that DaVinci Resolve plays much nicer with. Just, for whatever reason, the proxy media that was generated for the video files I was working with were something like 100 times larger than the source files, which is way too big for my local storage but no problem for my network storage. I didn’t try this solution originally because I thought DaVinci Resolve didn’t play nice with the source files because they were on a network share, and I never graduated from that line of thought while doing initial troubleshooting.
Anyways, I figured out generating proxy media was my solution so now I could slap videos into the timeline and edit cleanly with previews that kept up with the play head. We’re not quite finished with the difficulties yet, but the remaining hurdles don’t have to do with the technology anymore; the source of the remaining problems were me.
I feel like it comes across in the video quite clearly that I am not a presenter in the least. I sit in an odd position because I am actually very good at explaining things to people in a way that makes sense to them and allows them to take the idea and move forward with at least enough of an understanding to accomplish whatever goals they may have. This skill of mine is so sharp that at work, my colleagues will specifically invite me to meetings to act as that translation layer should confusion or misunderstanding creep into conversations. Me, somebody that would be classified as a developer, being invited to meetings for my soft skills and ease of interacting. The problem is that when working with a person, I can get a measure of how they think and what they can comprehend and why and tailor my explanation to that. As a presenter speaking to an audience, that relationship does not exist; you need to make assumptions about your audience and tailor your messaging with those in mind, a concept that is paralyzing to me.
There are a few reasons why it’s difficult for me to get around this problem; first is that I have little experience with true public speaking in general. In work meetings, my audience are colleagues of specific stripes, so it’s easy enough to guess their relative depth of knowledge and craft how I convey my messaging with those assumptions. But without experience in public speaking generally, I don’t have the tools or experience to determine what attributes I should assume about my audience or why. There’s also the problem that I don’t really have an audience right now, so I don’t even have real data to dig through to start forming a basis on either. Another limiting factor is that I heavily dislike the idea of tailoring a message to some wide audience instead of focusing it narrowly for individuals; really this can be seen as a personal failing in this case because it would really be best to be capable of both and understand when to apply each approach and why.
All this means that it’s difficult for me to formulate messaging in a cohesive way. And even if I could, I have another problem that makes negotiating all this difficult; my reading comprehension isn’t great, so doing something like writing down notes to organize my thoughts or writing up a script to read back actually makes recording more difficult as I can’t read fast enough to go back over the script in a natural sounding manner. If I have a loose outline of where I want to go and speak off the cuff, my presenting is at least more natural, but if I write up a script, while my thoughts are organized, reading through it is actually quite challenging requiring a lot of cuts and adjustments. There’s also the problem that I don’t have a proper DAW; I used Audacity because it’s free and simple, but I found that trying to redo a take would require some planning and editing that I wasn’t ready to try figuring out, so in the end I just worked from the cuff and I feel like it shows with how I meandered at times. For this specific project, while recording the audio was a struggle, I was more interested in getting something finished and posted because “done is better than perfect” as some rightfully assert. These problems aren’t impossible to overcome, but today was not the day to climb that mountain.
All of this brings us to the video embedded at the top of the post. The editing is simple, the commentary meanders at points, and there are no transitions to smooth out cuts, but it’s finished. I can move on and explore if it’s worth sharpening those presenter skills or focus more on editing other content rather than cutting together videos of my own, anything really. So now that you’ve endured reading about the technical struggles to get to even posting my first bit of original content, let’s get onto a far more interesting subject; that odd trade.
Art of the Deal
…or Deal for the Art I Guess
I don’t think I characterized the interaction particularly well in the video, however I’m also not about to go and try editing it further to get that context across. I think what I said in the video is accurate enough that most people will have an accurate enough picture of events, but since I have this forum where I can better organize my thoughts and present them in a more logical manner, let’s exploit those properties.
Part of the backdrop that’s probably not ultimately important to the narrative but is important to me personally is that I experience some level of Aphantasia; the shorthand is that my mind’s eye is terrible and I don’t really envision things so much as process thoughts as sensations and associated emotions. As such, when I “design” characters, I have an idea of who they should be, what kinds of personalities I think they should have, but I have no concept of how they should look and I especially have no concept of what about their looks would reinforce or otherwise point to those personality traits. Details like hair color, hair style, eye color, etc. never come to mind because they’re not related to personality, so I often outsource those elements to artists I commission when getting designs made. In the case of starting my content creation journey, I asked a friend to make that design for me, which is the current design I’m using for Amari Elder.
Just to add to the confusion, in the video, I noted this friend goes by Little Snow; that is technically true in Discord and I believe was the case on Twitter/X until recently I suppose unless maybe I’m inventing memories which is entirely possible. They now go by Lereleica on Twitter/X and Bluesky and some year I’ll refer to them by that instead of defaulting back to Liese, but we’re here for now (as an aside, you should check out Lereleica on BlueSky).
Now, Lereleica did me a massive favor when drawing up the original art I used as a PNG to stream and tried to set it up so that it could be rigged in the future as a model. Lereleica clearly had some knowledge of how to do this, but considering they’d not had art rigged before and I had seen how complicated rigging was, I suspected in order for the original art to be easy to work with when rigging, more work would probably have to be done. This was mostly just a note that existed in my mind because, at the time, I didn’t have any plans to get the model rigged because I didn’t have anywhere near the amount of money required for that.
Fast forward to probably a year and some change later, and I’d had the honor of getting to know Isa Lellebana. Now, in the video I really don’t do her justice for how sweet and fun a person she is, but I also think that it’s impossible to properly convey that; I say this, but here goes attempt 2 at trying to explain. In spending time in a specific corner of the VTuber sphere, I’d come across Isa and started watching her streams; one thing I know I didn’t convey right in the video is that she is an artist with a taste for creative endeavors she finds entertaining. To name a few of those endeavors, she draws her own models, rigs said models, knits clothes, does a bit of sewing, and has made a few video games. Also what became evident to me watching her streams and her interact with other streamers was that she tries to exercise kindness when she can. So when I caught her talking about doing low-effort rigging for cheap (a detail I misspoke about during the video, but wasn’t going to redo the take), I asked her to put a dollar price on it. Really I asked her for 2 reasons; one was because I was interested for reasons I’ll divulge in a minute, but the other was as a check to see if she would suggest an amount that’s fair for the required effort. I was satisfied with her response, even if I feel she was still underselling the effort required a bit.
So back to being personally interested in commissioning Isa to have my model rigged, there’s a few layers here. The most obvious incentive is that rigging is expensive generally, and I’m absolutely not in this space to create content, so the return on investment there for me personally is not going to be there, thus the less I can pay to undergo such a complicated process the better (there’s a pretty obvious conflict of priorities here as I want to be sure Isa gets compensated properly for her time and effort while also avoiding paying out the usually thousands of dollars rigging typically costs). Beyond that though, I still had that note in my mind that Lereleica’s art probably needed some help to be easier to rig, but given what I understood of Isa, she would probably be extremely helpful for my inexperienced artist friend. That meant the value proposition from my perspective was highly skewed in my favor; my artist friend gets experience to improve their craft and I get a rigged model at a relatively low price. So I made space in my budget for the rig and commissioned Isa.
Isa met my expectations; she pulled up the model on stream, took a look at the layers, provided feedback for Lereleica on improvements to make, provided resources on how to make those changes, then rigged the model on stream once the changes were made. So then came the hard part; I asked Isa how she would like me to pay her, and she hesitated to give me an answer.
This was a roadblock I had anticipated from the outset; in the limited time I’ve known Isa, she’d managed to strike me as the type of person that struggles to see the value of her own work, at least in strict dollar terms (hence why I was concerned with her undervaluing the effort she’d put into a “low effort” rigging). However, she was already comfortable with essentially me buying stuff for the value of the commission and having it sent to her, so she accepted the idea when I proposed it. One thing on the list was a LabM8 jacket; merchandise from AQUA191V (they’re awesome and you should check out their content). For what else to do to get up to the value of the commission, I suggested upgrading her computer. Problem there is that Isa is just a wee bit frugal, and by that I mean if she has something that works, she wants to essentially run it into the ground before upgrading to something else; honorable really and something I can understand as I have clothes that are over a decade old and riddled with holes at this point, yet I’m unable to just throw them out.
I was able to talk Isa into getting an upgrade for her current system in lieu of replacing her system because she was operating on a prebuilt HP machine that used an integrated GPU, which made streaming and playing games challenging, so she could logically understand that she needed an upgrade, but tossing the whole system in favor of a replacement was a bit much while doing some sort of in place upgrade was a lot more palatable. I got the model of machine she was using from her and looked up the variants and hardware they came with, and one variation of the model did come with a dedicated GPU; of course the HP model of the GPU was built as cheaply as possible with a single fan and I wasn’t going to get that for somebody that had been so helpful, so I ordered the EVGA model I’m seen working on in the video. We would discover once she received the GPU that HP was so cheap as to use 2 different power supplies with the 2 different models; a like 120 watt PSU for the base unit she had with a single CPU that came with an integrated GPU and a 400 watt PSU for the unit that came with a dedicated GPU, so I had to also order the more powerful PSU and send it to her as well.
Isa got the GPU and upgraded PSU installed and everything seems to be working just fine, and the LabM8 jacket should arrive when they’re made and shipped out so all in all, an excellent exchange really. From all of this, Lereleica got more experience crafting art in a way to make it easier to work with when rigging, Isa got an upgrade to her computer, Isa will get a LabM8 jacket, I got to order a LabM8 jacket to support AQUA (I would not have made this purchase on my own because I couldn’t justify it with my tight finances), and I got my model art rigged (one of these years I’ll finish setting up L2D studio and stream with it). So that’s the odd deal that ended up helping more people than intended; I love those kinds of interactions, mostly because it’s far more interesting than just handing off some cash and getting a product back.
Oh, and one last note to try clearing up what I was talking about with thermal interface material and GPUs; if you watched the video, there’s a point where I pull apart a PCB (printed circuit board) and a heat sink (the giant slab of metal with the fans on it); the goo looking stuff between the two is a kind of paste intended to allow the heat from the GPU die to be transferred into the metal to be dissipated by the fans on the other side. It’s imperative that the thermal interface material (a paste in this case) be in good condition to continue transferring that heat otherwise the die can overheat and cause internal damage to the die.
The typical advice to avoid this problem is for you to watch the GPU temperature, but there’s a few problems here. First off is that if you were to say open Task Manager on Windows, go to the Performance tab, and click on your GPU, you’re given one temperature that’s an average of something, and maybe not even something that’s really relevant to understanding the health of your GPU. Most modern GPUs have thermal probes (or other temperature measuring methods) built directly into the GPU die and on or around other components; these more accurate sensors can be viewed using tools like GPU-Z. This is naturally better as there are patterns you can use to determine if there are potential problems or if the thermal interface material is losing efficiency, but it’s also not an infallible tool or method by any means. What I was trying to stress in the video is that it’s better to simply preventatively have the maintenance done, say by taking your GPU to a local shop if you’re uncomfortable doing the work, than it is to try a strategy like installing an external tool and those temperature measurements to determine if you might have a problem. The weakness in this strategy is that installing probes is additional cost and complexity to a product, so while there are a lot of points of measurement for temperature, specifically so the GPU driver can control the card and keep it operating within its design tolerances, there aren’t probes everywhere, and it’s entirely possible that the thermal interface material dries out in such a manner that this unmetered space gets subjected to excessive heat before the symptoms manifest on any reports from those third party tools. If you want to be pretty well guaranteed to be safe, doing that maintenance every 2 years is probably a solid bet. For most people, I’d say getting the maintenance done at the 4 year mark is probably fine. If you’re getting to the 6 year mark and haven’t had your GPU thermal interface material refreshed, you’re playing with fire and should get that done as soon as possible if you want to avoid an expensive component failure.
Anyways, as a fun bonus detail to the nerds out there reading this, the material I used on Isa’s 1650, and why I said it should outlast the remaining useful life of the card, was PTM7950; it’s a phase change material very different from traditional thermal paste. Its greatest strength is that it doesn’t have the dry-out problem that thermal paste does, meaning it has a much longer service life. It’s not exactly cheap, but neither is a new GPU.