As I'm writing this I haven't even uploaded part one and part two yet. I’m taking Gear Talk seriously, literally no one else is talking about what it takes to make a real non-fungible non-social media, non-institution-backed project. The truth is it takes very little to start, but to self-sustain is near impossible, expand? Pretty sure I am the only one crazy enough to try. Then I remember Linus Sebastian’s videos of when he dumped all that money into creating Linus Media Group, before anyone took YouTube seriously, and it becomes reassuring. No one knows what’s next — apart from those building for it, because it’s them that will make it a reality. I’ve finished up the cameras section, and debating if you need pretty pictures for it to hold your attention, and in terms of workstation this is the definitive lesson. It’s been a long few weeks of trial-and-error, and the last thing I want is to have to rebuild a system that is packed to the gills with project files, so starting over now is relatively painless in comparison to mid-project. The other two parts are important to see how I got to this conclusion, but ultimately this is where the situation is at.
Firstly, Nvidia is basically halting production of the middle-segment of cards. Some insiders have confirmed this, official statements say otherwise. In my experience no brand ever announces discontinuation, product just disappears. If you are a gamer that means you play games at a slightly lower setting, simple enough. As a creator? You decide if this is the peak of the market, or another Covid situation. I believe its another market collapse, the middle gets gutted, and inflation makes both other and upper tier products unviable. If we are lucky we get a “New Coke” scenario, and when “old Coke” comes back it is multiples in price. Worst case we get the “chocolate bar” effect — bars get smaller as prices get higher. I decided one simple fact early on, you cannot be just a “photographer” in today’s market. I decided I wasn’t going to be a marketer, or a videographer, but a project manager — one in a juvenile, mostly inflated, domestic market at that. My choice to buy the 5070 and 5080 are largely proven correct, which is good as I could have bought a lot of other things with that money; from prototypes for Discover Temples of Thailand, a week’s scouting in Japan, a 50mm f/1.2 G Master, or just more production time.
However, as relevant as all of those things would be to making the work, my workstations are what is relevant to keeping the work. It isn’t the most flexible of configurations, but it is immensely more flexible than what i had prior, meaning more decisions can be made in post-production. Not as most describe “fixing it in post” but flexibility in direction of the deliverable. When you are “hard locked” on time, or sometimes just efficiency (largely two sides of the same coin), the only option you have is to iterate. If I was making this from the West, a follow-up would be immensely expensive in travel costs, pick-ups would be prohibitive. A domestic artist could not survive with this outlay, and that is to say the window of opportunity I have built over the course of the past fifteen or so years has put me in a singular position. Hardly a position of abundance, but one of flexibility and control of my horizon. If anyone is reading this in the portfolio procurement stage, scrap any of this as actionable, that week in Japan (or wherever) is your priority.
This may be a radical stance in 2026, but I think imaging has regressed to superficialities. For the most part, I believe that once the camera and lens part of the equation is resolved (pun slightly intended), and yes it can be resolved, capture becomes a miniscule part of photography. The term “Digital Darkroom” is largely a misnomer, as I think it is far more than that, and in some ways far less. I do think we have reached a point where AI interpolation has become good enough to constitute as a successor to enlargement, but I (and my school of Philosophic Documentarianism) believes that a photograph is the data captured by the sensor, and anything more begins to take it from the realm of photography into digital manipulation. I believe what I do in particular to be authentic, with monochrome conversions a selection of wavelengths captured by the sensor, an interpretation of the data not alteration of the image. I largely and publicly disagree that the image itself is a product solely of the camera, and a strong advocate of ethical post-production. Otherwise, wouldn’t we all be still shooting positives?
Yamato & Musashi
I’m not usually one to name my things, I have never named a car, a bike, or a camera, but I felt the need to name my two computers. Initially inspired by Space Battleship Yamato, released in 1974 (and a whole five years prior to Tomino-sensei’s Mobile Suit Gundam), it is a story of how in a last ditch attempt some scrappy folks on Earth repurposed the naval battleship Yamato for intergalactic travel. It seemed fitting, I was asking my computer to do something it was never intended to do, and I was upgrading it in a time where [generally] nothing is available, rather fitting. At least until I decided to build out the second, and the only name that was suitable was obviously Musashi, the sister-dreadnaught. What becomes eerily prophetic is how both builds mirror the production of the original vessels so closely, it gives me goosebumps. Musashi was always meant for reserve duties, and Yamato the flagship, the bearer of the Rising Sun, dominator of the Pacific. What happened was Yamato, by virtue of being first, was restricted in it’s construction. In all accounts it was the flagship intended, but Musashi was able to learn from Yamato, retrofitted and adaptable.
Let us take a refresher on what system I was using between early 2023 and late 2025:
Motherboard: Asus TUF Z790-P WIFI DDR4
CPU: Intel i7 13700
CPU Cooler: Cooler Master 240mm AIO
GPU: Asus TUF 3060 12GB
RAM: Corsair Vengance 32GB 3200 (2x16GB)
SSD: 1TB WD Black SN770 x2
PSU: Thermaltake Tough Power 850W Gold (Modular)
Cooling: Thermaltake 120mm x3 (intake) & 120mm (exhaust)
Chassis: Thermaltake V250
This is not going to be a spec list buy this article, or a product review. It will be something much more useful, transparency. This is the setup that I was using for a few different things initially, mainly because transfers were becoming impossible on the MacBook Pro, even on a normal 35c day here in Thailand, the machine would overheat. It meant any post-hoc backing up was virtually impossible, data management, archive management, not hard, not stressful, but impossible. This was post pandemic shortages Udon Thani, in a country where returns don’t exist, the mail is the wild west and you can forget importing anything. Depending on how you look at it, I was lucky that I got anything at all, or short-sighted in that I didn’t optimise. Either way, speculation remains unhelpful as the truth is it was the system that I had, the system that was available, and more importantly; the system that I used to produce multiple projects, maintain DRobertsPhoto.com, and use through my editorial contributions (that number over twenty, with seven optical manufacturer collaborations). This is Yamato, named recently when repurposed to handle Discover, and both the data-vault of 60 megapixel RAW files that the project is producing along with H.265 4:2:2 motion companion pieces — and separate experimentation in extremely costly (as in performative cost, not financial) applications.
These were the two workstations as I entered into 2026:
Yamato
Motherboard: Asus TUF Z790-Plus WIFI DDR4
CPU: Intel i7 13700 (-0.07 offset)
CPU Cooler: Corsair Nautilus 240
GPU: Asus TUF 5070 12GB
RAM: Corsair Vengance 64GB 3200 (4x16GB)
SSD: 1TB WD Black SN770 x2 2TB WD Black SN850X x2
PSU: Be Quiet! 1000W Gold (Modular)
Cooling: Be Quiet! Pure Wings 120mm x3 (intake) & 140mm x1 (exhaust)
Chassis: BeQuiet! Pure Base 501
Musashi:
Motherboard: Asus TUF B760M-Plus WIFI DDR5
CPU: Intel i5 12400
CPU Cooler: Be Quiet! Pure Rock Pro 3
GPU: Asus TUF 3060 12GB
RAM: Corsair 32GB 6000 DDR5
SSD: 2TB Kingston NV2
PSU: Thermaltake Tough Power 850W Gold (Modular)
Cooling: Thermaltake 120mm x3 (intake) & 120mm (exhaust)
Chassis: Thermaltake V250
With Dararat taking the lead on the Dehancer collaboration (see here), it is clear that DRobertsPhoto is growing in the ways that I want it to. To support and nurture this, like a plant, it needs to be watered. I could have easily just built a singular “hero” system for my own use, and come in at around the same cost but the idea was to iterate enough that Yamato would be in fighting condition, and what was left over could be supplemented by a comparatively small purchase to create a second workstation. In theory it was a good idea. In practice, a capable low-cost machine was built if working from proxies. In use? A headache. Proxies are industry standard for motion, that much is undisputable but the only real benefit that comes with independence is agility, and proxies add friction between capture and edit. Motion is still in the early stages and heavily reliant on iteration, and when things become difficult the willingness to try disappears. Iteration becomes a chore, and it becomes toxic for future development.
And how they look today:
Yamato
Motherboard: Asus TUF Z790-Plus WIFI DDR4
CPU: Intel i7 13700 (ASUS OC 90c limit)
CPU Frame: Thermal Grizzly
CPU Cooler: Corsair Nautilus 240
GPU: Asus TUF 5070 12GB
RAM: Corsair Vengance 64GB 3200 (4x16GB)
SSD: 1TB WD Black SN770 (OS) 2TB WD Black SN850X x2
PSU: Be Quiet! 1000W Gold (Modular)
Cooling: Be Quiet! Pure Wings 120mm x3 (intake) & 140mm x1 (exhaust)
Chassis: BeQuiet! Pure Base 501
Yamato hits a consistent 28,000 (+/-) on Cinebench R23 at 90 degrees (100% load, 4.9GHz +/-) that was reduced 3-5 degrees with the Thermal Grizzly Gcontact frame and Duronaut application. Idles at 40-45 Degrees, generally under 60 at load, 75-80 turbo with 30 degree (+/-) ambient
Musashi:
Motherboard: Asus TUF B760M-Plus WIFI DDR5
CPU: Intel i7 14700 (ASUS OC 90c limit)
CPU Frame: Thermal Grizzly
CPU Cooler: Corsair Nautilus 360
GPU: Asus TUF 5080 16GB
RAM: Corsair 32GB 6000 DDR5
SSD: 1TB WD Black SN770 (OS)
PSU: Corsair RM1000e 1000W Platinum (Modular)
Cooling: Be Quiet! Pure Wings 120mm x2 (intake) & 120mm x1 (exhaust)
Chassis: Phanteks XT M3 Air
Musashi hits a consistent 34,000 (+/-) on Cinebench R23 at 80 degrees (100% load, 5GHz +/-). Idles at 30 degrees (+/-), generally under 55 at load, similar 75-80 turbo spike to Yamato with 30 degree (+/-) ambient.
Now if you read part two, you know this was not a clean iteration and I wasted more than I should have getting here, in understandable terms: if I did this off the bat, I could have had 64GB in both systems and dual 2TB 850Xs, but some lessons are learned the hard way, others are learned the expensive way. I have a pair of SN850X in 1TB sat on my desk, with the plan of replacing the OS drives, and I am still contemplating a pair of 2TB SN850X specifically for Musashi, but the NVMe drive slots on the B760M are right behind the GPU, a GPU that just so happens to be the 5080 now. The logical thing to do would be to just put the 2x 1TB drives in for media and call it a day, but I think I have gone past just building a workstation and have begun optimising in the same manner that I have my cameras — where the bell curve of price to performance is largely ignored so that I can insensibly domicile myself within the worst part of the graph.
I don’t think this is my final iteration, but I also hear my own words calling back to me. What I could spend next could be a trip to Krabi or Pattaya. Not just a change of scenery for my work, but for us personally. This is a sensible place to stop, and both systems function well. Any increase in performance is now marginal gains rather than the proxy/no-proxy choices of before. It’s a little emotional as well, Yamato has served DRobertsPhoto in the worst of “battles” this past three years, and part of me does not want to give it up. A part of me wants to re-case (again) so that I can get a 360mm radiator in there, so that I can replace the 13700 with an i9 14900 and pair it with the 5080. I even have a case in mind, the Havn BF360 with enormous 180mm front fans and an air-duct that would feed my NVMEs directly (on the cool underside of the GPU). That would also solve almost all the niggles I have with Musashi as well, the 5070 being a far tamer card for heat management due to the proximity of these NVMe drives in that system. In a way, I shot myself in the foot trying to maximise gain for minimal spend. In another way, this upgrade today would already cost 10-15% more than it did only two weeks ago, that is the rate of the absurdity we are seeing within the market. In the cleanest of ways to justify this now; I am lucky to have been able to build this at all due to stock. Could I have built a DDR5 5080 i5 14700 setup? Sure, there is always something available, but I would have been buying budget-focused product SKUs (such as the plentiful i7 14700F or ASUS Prime offerings), or hyper inflated gamer/creator focused equipment (Such as ROG or Pro Art). ASUS TUF is the gold-standard for OC and thermals and virtually no “cool tax” penalty. WD Black 850X are end-of life no matter what Sn Disk may say, and I am hoovering up what is available locally. As someone pointed out; inflation here is relative to the demand of a domestic market, however when product lines are sold out it will inevitably become a full system collapse because the lower demand will no longer be facilitated. Companies, suppliers, distributors will prioritise the never-ending 300% markup of the US, or EU over SEA. When Apple stock of memory reaches zero? When 2025 SKUs come end of life? Do you believe 2026 model will not be inflated due to NAND pricing? Even with Apples purchasing power, and seeming disinterest in competing in enterprise-AI it inevitably will be. We are in a hostile time to create, economically, ethically, educationally.
If I were to pursue a “final” iteration it would look something like this:
Yamato
Motherboard: Asus TUF Z790-Plus WIFI DDR4
CPU: Intel i9 14900 (ASUS OC 90c limit) - Projected 17,500 THB cost, 38,000 R23 Benchmark
CPU Frame: Thermal Grizzly
CPU Cooler: Corsair Nautilus 360 - Projected cost 4,500 THB
GPU: Asus TUF 5080 16GB
RAM: Corsair Vengance 64GB 3200 (4x16GB)
SSD: 1TB WD Black SN850X (OS) 2TB WD Black SN850X x2
PSU: Corsair RM1000e 1000W Platinum (Modular)
Cooling: Havn 180mm x2 (intake) & 140mm x1 (exhaust)
Chassis: Havn BF360 Flow
Musashi:
Motherboard: Asus TUF B760M-Plus WIFI DDR5
CPU: Intel i7 14700 (ASUS OC 90c limit)
CPU Frame: Thermal Grizzly
CPU Cooler: Corsair Nautilus 360
GPU: Asus TUF 5070 12GB
RAM: Corsair 32GB 6000 DDR5
SSD: 1TB WD Black SN850X (OS) 2TB WD Black SN850X - Projected cost 10,000 THB
PSU: Be Quiet! 1000W Gold (Modular)
Cooling: Be Quiet! Pure Wings 120mm x2 (intake) & 120mm x1 (exhaust)
Chassis: Phanteks XT M3 Air
In a way the modularity that comes with building on the same TUF/50-Series/LGA-17000/Gen4 PCI-e framework is what allows this experimentation and iteration between systems. If I were to go Mac, the only option would be to buy a new system. If I was min/max performance deltas then I would be torn between AMD/Intel.
As I said multiple times in these Gear Talk segments; the question isn’t is it worth 30,000 THB anymore, but “Is it worth Krabi, Pattaya, The Sanctuary of Truth, Korat, Chiang Mai, or whatever else I could do with that instead?” With The Nakasendo Way planned, and expansion into Japan paramount subject is not lacking, neither is materials for Discover Temples of Thailand locally; but neiother is the optionality black and white. The upgrades purchased where clear cut; proxies or raw editing on system two. That is no longer the case (pun not intended). The case is whether I want Yamato to lead, or Yamato to follow. Do I maximise the potential of the LGA 1700 socket in a period of “chocolate bar” economics (by purchasing the slab of Hershey’s even at extortionate rates before it becomes sample-sized), or do I say enough is enough. Unfortunately art does not exist in the logical domain and the Project Manager does not win all battles. Perhaps I should say fortunately, as that’s my external battle, isn’t it? I think the one thing that I have taken from this experience is that I knew from the beginning that a 5070 or an i5 was a compromise, but I had hoped it was not. That costed me real money but taught me (or perhaps reminded me) that compromise costs more than commitment does.