I’m sorry, this is off-topic, but I just saw this on Twitter and I don’t think I can rely on hamsters that can grow random arms 
Adding even more arms might make it look intentional at least…
7 Likes
Hahahaha ! looks defenitly intentional, as it’s an analogy of choosing a render engine, you can’t have both, like you can’t have three arm, each one has its advantage but eventually you have to pick one…
These kind of subliminal messages really helps ton convey the message !
4 Likes
This hamster just was very exited and so “waving” the arm very fast… the “camera” just was not quick enough to “follow”..
3 Likes
“All the Hoopy Froods use Farpy!”
The marketing practically writes itself. 
1 Like
Yeah you do you. Thanks for engaging.
1 Like
Farpy is a great name, It’s so bad its great.
1 Like
I render my interior visualizations set up as animations.
Are you prepared to do 30 hours of rendering for 30 cents? That’s awesome. 
But if you were, that would raise questions… Why? How does that pay off? I would probably just come to the conclusion that you are after assets in my files.
4 Likes
Fair question! Thx
The goal isn’t to do 30 hours of rendering for 30 cents forever.
The idea is to use spare GPU capacity that would otherwise be sitting idle and keep pricing simple enough that people will actually try it.
As for assets, I’d expect people to be skeptical of any new render service. Trust has to be earned over time, not claimed. Hmm, if someone has highly sensitive studio work, I’d expect them to stick with providers they’ve already vetted. That’s completely reasonable.
Right now I’m mostly focused on proving a boring workflow:
upload .blend → render → download ZIP → receipt
and then letting trust build from successful renders rather than marketing claims.
1 Like
The fact that you advertise RTX 3090s being used also does not inspire trust. Why an old consumer GPU?.. If it was a serious render farm, wouldn’t power efficiency be a concern? Also what’s up with failed renders? If I need to render something, I need it not to fail. Right now it seems as if you are saying, “Well, maybe you will get the result, maybe not, but if not, don’t worry, you don’t need to pay then.” I think people pay renderfarms for certainty as well. Of course the clients need to make sure their files do render.
“Failed renders cost $0” - that’s a very dangerous claim. Anyone who has any customer service experience will know 
What’s a failed render? If I make a mistake in the file, and I consider it failed, but you rendered for 5 days anyway?.. I think you should have an extremely clear and specific definition of what “failed” means here.
5 Likes
Fair.
The 3090 thing is actually me being transparent. I’d rather tell people exactly what’s rendering their file than hide it behind marketing fluff.
As for “failed renders cost $0”, I agree the definition needs to be crystal clear.
To me, failed means:
- render crashed
- job never completed
- output couldn’t be delivered
- system-side problem
Not:
- customer changed their mind
- scene had a mistake
- wrong camera
- bad lighting
- forgot a texture
If the render completed and produced the requested output, that’s not a failed render.
I think reliability is the bigger challenge honestly. Price is easy. Trust is earned.
1 Like
I would put that into some sort of terms and conditions maybe, not the front. I mean for me it would be pretty obvious that if there is a problem with the render process at the render farm they would not charge me for services that were not provided… I think that’s pretty standard with anything. Specifically mentioning it may just direct the attention of possible clients to the possibility of services not working as intended without a good reason.
3 Likes
I’m not going to argue with you about it, it’s your project, just know that I would never consider using it unless the name changed. No one in a professional setting is going to write a memo suggesting a trial of something called Farpy
1 Like
I’d probably try it for small test renders or quick previews, especially if pricing really stays that low per frame.
What would make me hesitate is reliability and trust. Blender renders are super sensitive, so I’d want clear proof the output matches local render results, plus some way to preview before committing a full job.
Also the “failed renders cost $0” is good, but I’d still worry about repeat failures or hidden limits. If it’s consistent and actually faster than setting up my own render queue, I can see the use case.
2 Likes
you can usually contact the farm and they can arrange that for you.
Yes you can. The upload managers of the render farms I have tried handle that perfectly fine.
All of that is taken care of by the upload managers.
Now, that doesn not mean that farms are a smooth experience. large caches can be limiting. If you don’t have a very fast internet connection then uploading 500GB can take quite long.
Some upload managers can check if your files have changed since the last upload. So if you upload another iteration of the render file it can detect if the 500GB cache has changed and will then not have to upload it again.
Another annoyance is that render farms seem to have to give the ouput files of your image sequence silly names instad of the one you define in the blend file. You have to batch rename them after downloading. Not a deal breaker but it is something that tends to get on my nerves when the deadline is ticking.
So, I really prefer to render locally over night or weekends but with short deadlines render farms can make sense.
3 Likes
Yeah you right that many problems can be solved, but there is still inherent problems vs a local renderfarm. To my knowledge there isn’t a system that really emulate that, like you just have to hit render and images starts to pop in the network as they are rendered.
In this case while you render faster, it also creates some kind of a latency and it add file management.
I think then it goes down to what kind of work you’re doing. In mine shots tend to render fast, there is a lot of them like maybe 150mn of content to produce on the average project but this can go to 500mn over 8months / 1 year.
Keep in mind that 500mn means rendered several time for each departement + all the retakes.
Online renderfarm might sounds mandatory for that, but given the dynamic explained earlier it’s possible to render stuff locally without being dragged down. And generally it makes things simpler to manage…
2 Likes
Nah, probably not.
I guess one could create a custom script for rendering on AWS or similar cloud cumputing. Something similar to what the Brenda scripts did back in the days only with some auto download into a the folder specified inside the Blender file.
1 Like
That’s fair. If the product works, the name becomes normal. If the product doesn’t work, the best name in the world won’t save it.
2 Likes
That’s exactly the use case I’m trying to prove first: small test renders, quick previews, and short frame ranges before anyone trusts it with a full job.
Reliability/trust is the right hesitation. The current direction is: preview first, receipt/proof after completion, failed renders cost $0, and speed tiers only when they create real wall-clock improvement.
Hidden limits are fair too. I’d rather make the limits obvious than pretend every Blender file will work perfectly. The goal is simple: upload .blend, render faster than local, download ZIP, no babysitting.
1 Like
That’s fair, and honestly that’s why I’m starting small.
I’m not trying to convince people to abandon local rendering. If you can render overnight on your own machine, that’s often the simplest option.
The opportunity I see is the gap between “render locally” and “manage a traditional render farm.” Upload a .blend, render, download a ZIP, done.
A lot of the friction you mentioned is exactly the stuff I’m paying attention to: assets, caches, file naming, repeat uploads, proof that the output matches expectations, and reducing the amount of babysitting required.
For me the success case isn’t replacing every render workflow. It’s the moment where someone has a deadline, their local machine isn’t fast enough, and Farpy is the easier option.
I think that’s the interesting part. Most people don’t want to manage cloud rendering, they just want Blender to finish faster without changing their workflow.
1 Like