I keep seeing Blender users needing a few frames rendered but not wanting another monthly subscription.
So I built a small render utility:
Failed renders cost $0.
Pricing starts around 1 cent per frame.
I’m looking for honest feedback:
• Would you use something like this?
• What would stop you from trying it?
• What would make you trust it?
Doesn’t a lot of render farms already work in this model, of paying per render (either still images or video)?
Edit: other than that, I don’t know the economics, as I never used a service like this. So, you probably should explain what benefits your service offer, when compared to similar ones (speed? reliability? cost? easy payment?)
I would be instantly dubious of being charged per frame, not based on the amount of time. Since one frame could take 20 minutes to render, which will draw significantly more power than one frame that takes 30 seconds. Yet from what you have written it sounds like it would cost the same.
I also don’t really trust any small render farm, I want one that has been around years and trust. I also preferably would want them to have some kind of certification to ensure proper security, such as ISO 27001.
With something like what you are talking about, I have the same issue that I have with SheepIT, I don’t really know where my data is going, and the price seems to good to be true.
The information about privacy, security, and terms of service seems pretty vague on the website you linked, which further lowers my trust.
I am now saying you are doing anything bad or untrustworthy, I am just saying personally I wouldn’t feel comfortable submitting my data to it
No, I wouldn’t.
I have a 3060ti. My frames take under 7-10 seconds to render, and require using paid addons that your system won’t have installed. Further, I’d have to bake a lot of the animated content so that it would not have intra-frame desync due to multiple machines evaluating/rendering frames out of sequence. Then I also have to download hundreds of frames from the server when complete.
All of that = it would be more work for me to use such a service, and cost me more time that just rendering it myself.
Well, personally I don’t use renderfarm services because IMO they are a pain to use. But in some cases it could come handy…
Charging per frame ? why not, at the end of the day, the main reason one would go trough all the hassle of using those services is to save time.
For instance if I have a 2mn film , at 5mn per frame I might be looking at 250 hours of render for the whole : (120 seconds * 25 images * 5mn) / 60
Generally it’s something we would do over the course of the project.
Indeed if I have 50 hours of renders to do, that I could cut down to 1h with a renderfarm services that might be worth it.
In fact, 50 hours might be still doable locally, because I can start working on some shots, launch renders and the client might be willing to wait two days.
Maybe if we talk about 100 hours brought down to ~2/4 hours it starts to make a real impact on the project.
Now the way it’s invoiced for the renderfarm service, really doesn’t matter, it’s the end bill that counts. And price can quickly jump.
If we talk about your service : that 2mn film would cost 150$, which is pretty reasonable to include in a client’s budget.
Problem is , it’s not about rendering once and bob’s your uncle : client is likely to ask for change, I’ll very likely make some mistakes so while the render might be valid, I’ll have to render again which is pretty common. So that initial 150$ bill can quickly turn into 500$ or more. which starts to weight in the balance, but well, maybe it’s a premium project that asks for very responsive feedback with client.
Then there is a lot to say about why in such a project using an online renderfarm is a pain. And it’s in two parts : Some stuff you can’t fix because it’s inherent to that system, and some stuff that depends on how the service is designed…
Some stuff you can’t fix because it’s inherent to that system :
you can’t use addons,
you generally can’t link libraries or use stuff like vdb, caches…
it fits better a product viz / arch viz / mograph project that an animation project that relies on external assets…
each time you want to render a shot , you need to : pack texture, upload, extract renders and put them in the project folder so it’s accessible for compositing. ( generally we don’t want filepath to change). Anyway, this is fine if you do that only a couple of time, but for a 2mn sequence we might be looking at say 30/100 shots x ~5 re-render , so it’s a lot of manual work.
some stuff that depends on how the service is designed :
TBH it’s a long list : it’s a lot of work to make an production ready renderfarm submitter.
Stuff like choosing blender version, rendering in exr multilayer…
Is python enabled ? if yes I might be able to hack the service, if not stuff like drivers will break. Does the web interface allows to set a specific time range (like frame 50-> 150)? And a lot of detail like that.
You need to work closely with some artists to make sure they can all use the tool given their needs. Which is some kind of a chicken and egg problem. Why would someone spend a lot of time in counselling and testing to get a functional plugin, while they can go right through to a more professional service like ranch computing ?
What these service like Ranch do is that their hire artists and TD to build the product and do test case scenario projects. I’m speaking about Ranch because I’ve met people working there. But I suspect it’s very similar in other renderfarm services.
Finally, these kind of ~2mn product viz project I’ve done a few of them, and to me I prefer a lot to render overnight and on the weekends and being able to cancel a render as soon as there is an issue, using all the needed addons, and having renders that always go to the right folder to be loaded automatically in comp later.
If I’m working on a studio they generally have a local renderfarm so each computer can be put to use, which helps a lot to make render manageable.
Our previous 50h render example can for instance be divided by 4 for a small team’s studio, which turns a weekend stretch into just an overnight wait which is generally fine…
Fun fact is, that I’m currently using a renderfarm at home with only my workstation to be able to manage renders. So I can batch several shots, re-render parts and so on…
Anyway, that’s already a lot of feedback, I can probably elaborate more on some aspects… I think it’s important to hear about people like me that are really not found of online renderfarms, but obviously some people do use them, they probably manage to balance that more in their workflow and probably aren’t going to have as much strong point againt that than I do…
Feel free to ask if you have further questions !
That’s a really key point. If a renderfarm has - let’s say, 2-4 GPUs available - there’s just not a lot of significant time/cost saving to be had, compared to just buying the cards myself and setting up a farm in my office. Which is easily done, with server/client software I already have.
If a farm has 200 GPUs, then of course the math changes significantly.
Yes definitely ! If it’s your business it’s totally worth it ! After 6 months / a year you’ll probably be saving money already !
And TBH it’s super rare that you really need to render in super fast mode.
With 4 PC it’s already manageable.
I think I’ve seen a renderfarm service used once, which was basically that a ~2 months project which was partially erased which needed to be re-render very close to the deadline.
I don’t recall exactly how they managed… but it’s likely that they used a mix of local renderfarm with their ~10 workstations and using renderfarm services for the heaviest shots… Which wasn’t simple because shots files weren’t thought for using renderfarm services…
Anyway for a small studio 10PC + nights and weekends is generally fine for nearly all projects, even on short deadline which might benefit something like Eevee which is also something that makes these renderfarm services less useful.
This is also why these services support multiple software to broaden their offers… Because it’s not that useful in the end…
The issue I have is charging per frame not based on compute time, since as I mentioned earlier, some projects might take secconds per frame, whilst other minutes. This means it is costing the render farm more but you are paying the same, which makes me dubious as to how they can survive.
Yeah they may want to revisit that part later… I think it’s part of the development / creative process !
Fun fact, I met Sheep-it main developer when they just started their renderfarm project. We became good friends and I followed the first iterations of the project and eventually how it became sheep-it, a few years after !
And I can guarantee that there was a lot of iterations over the functionalities and fine tuning that to become something that could genuinely work. They made huge effort to understand how a 3D project works, what artist would need and carefully crafting something that initially looked like a proof of concept really hard to use in practice, into something that can adapt to many use case scenarios.
I’m sure here it’s going to be a bit similar if they want to keep running into that rabbit hole … Pricing, UI/UX, servers, nodes there is a lot to study and account for !
“(Even) One cent per frame” would be prohibitively costly. I have no problems at all with subscribing to a render-computing service, but the pricing model needs to be simple. And, not so expensive as to persuade me not to use it.
I recognize that high-powered computing resources are (very …) expensive to buy and maintain, and that the costs thereof must be amortized fairly. So, I’m perfectly willing “in principle” to pay money to those who have invested in those resources, which I could never afford to (and, would not need to …) buy myself. But the pricing model has to work.
Idly thinking: One possibility would be a model that allows me to “submit a job.” The costs of that job are calculated and presented to me in advance. I expect you to accurately tell me by which time I can expect the job to be completed. When the job does run correctly to completion, “correctness” being your correct execution of my specifications for it (even if I screwed up …), then I am responsible for paying. (I should be able to see the results, and “bill me at the end of the month.” You extend to me a certain amount of monthly credit to fairly protect your interests.)
Bonus points if you offer: “Run this job for the first [ten] seconds.”
Maybe I am given a certain allowance for screw-ups, but then again maybe not: I’m expected by now to know what I am doing. But, I should know how much money I’m about to spend, and to pre-authorize the spending of it for future billing.
Really? That sounds to me like the lowest price ever offered for rendering.
1200 HD frames takes an hour to render: $12 USD.
1200 8K frames takes a month to render: $12 USD.
That’s a freaking bargain. James Cameron would love those rates.
That’s exactly what I was thinking, you could render extremely resource heavy frames with loads of samples, and it would cost the same as a quick render with a few samples and just a few objects.
@thorn: I’m just talking “in principle.” I personally have never purchased such services, and therefore do not know the pricing. I’m simply describing how one might approach it. And, I guess that I really don’t know what I am saying. 
Well, apart from the waiting a month to see anything, he would hate that bit.
Which in one way is another big factor. I suspect for many to use a render farm, time likely is a pretty big factor. As in, you have a deadline, the job needs to be done as reasonably (price wise) fast as possible.
If you can wait a month, etc then outside of just not having the hardware to render it at all, one may as well just render it locally.
Yep, plenty do. My goal is less about inventing a new pricing model and more about reducing friction.
Upload .blend → render → ZIP.
No GPU shopping, no render-minute math, failed renders cost $0.
Still early, so I’m figuring out what users actually value most.
That’s completely fair.
For a production studio I’d probably make the same decision today.
Farpy is still early, so I don’t expect people to immediately trust a newer service the same way they’d trust a provider that’s been operating for years.
The frame-based pricing is mainly an attempt to keep costs predictable and simple for users, but you’re right that render complexity varies dramatically between scenes.
The trust point is probably the bigger challenge. Things like longevity, security practices, clear terms, privacy guarantees, and a track record of successful renders matter a lot more than pricing alone.
Appreciate the feedback. Comments like this are useful because they highlight where trust still needs to be earned rather than assumed.
This is excellent feedback.
A lot of what you described is exactly why I’m trying to keep the first version extremely simple and aimed at smaller Blender users rather than competing head-on with established studio-focused farms.
The biggest thing I took away from your comment is that the real cost isn’t just rendering, it’s the workflow around rendering. Uploading, packing assets, handling revisions, managing outputs, dealing with addons, caches, version mismatches, and all the little production details.
For many artists, especially on larger projects, that’s a bigger problem than the render time itself.
I also agree that trust matters more than pricing. A cheap render is useless if people don’t trust where their files are going or whether the workflow will break.
Really appreciate you taking the time to write all of this. It’s probably one of the most useful pieces of feedback I’ve received because it comes from someone who actively avoids online render farms and explains why.
One question:
If a render service solved only one thing for you, what would it be?
Faster turnaround?
Less workflow friction?
Better addon/library compatibility?
Or simply trust and reliability?
Thx again
I think for my workflow, which doesn’t involve many add-ons and has packaged assets already, trust and price are the main deciding factors.
I want something that is trustworthy, clear terms and privacy policy, ISO 27001 certified, and knowing my data remains with a trusted provider. As I am not a professional though, I also want a balance in cost.
I find GarageFarm a great provider for this, they have been around for years and offer the pricing I like, whilst also being trustworthy. I also like how they give $25 credits no credit card required just for signing up (I did most of my rendering with them on these free credits!)
I haven’t used render farms in a while, since I primarily used them when I only had a laptop, now I have a desktop with a GPU, I render on that.
I would only use a render farm if the project was extremely heavy and couldn’t be rendered on my own hardware.
Hello !
Cool ! I’m glad this is helpful !
I think the most important is being able to render fast, since it’s the point of using a tool.
On a professional project it’s unlikely to render everything with external renderfarm, because of the file management, but maybe for a few challenging shots it might come handy.
Supporting multiple blender version is good to have too, since we don’t use all the latest and it’s probably not the most difficult stuff to solve.
Yes ! I think it might be better to support simpler projects. For small studios, there is probably more money to be spend but they’ll have a lot of constraints in the meantime. And for bigger studios they tend to use all the PC at their disposal.
Another thing that can be useful is to be able to submit a .zip alongside a .blend. In the zipfile you’ll have a .blend and some extra files , like textures, caches, or any dependency that can’t be packed into the .blend.
Not super important but it’s part of small details that will make your system more useful.
Good luck !
My only feedback is on your name, which is, to be candid, awful. “Farpy” doesn’t communicate anything about a render farm. It reads as Far Py, I would assume it was a python library having something to do with distance from the domain. As one word, it feels like a Pokemon reject, and it is dangerously close to “farty”, or even “fatty”. If I went to my boss and said “hey lets use the new farpy render farm”, he would hear “farty” and would call me Mr. Farty for the rest of my career