I started using Blender a couple of weeks ago. My computer is not a hyperbeast of a machine, but it’s okay. The thing is that I tried making a grass animation from a Youtube tutorial, and the render time suggested by the program was 1,5 hour. To me that’s nuts, since I’d rather spend time modeling and sculpting. I’ve been making electronic music and rendering a song doesn’t take very long in programs like Ableton.
This is probably a very well discussed topic already, but I am just searching for clarity on this subject, because I think this 3D thing is very cool, and something I’d like to explore more. Finally we get to the questioning. I tried workarounds, but the render time stayed the same or even got worse. I know some of you may have external graphic cards, and very good machines. So are these render times normal? What is the optimal gear for a newbie? What do I do? :yes:;):rolleyes:
Animations usually takes much longer than that, some of my realistic images take about 40,60 min to render, since an animation is nothing more than a sequence of frames and one second takes 24 frames, you can do the math to estimate how long would take to render an animation of 1 minute… And by the way, particles takes even longer to render…
You should expect rendering to take long periods of time. Rendering animations can take months, not hours. I am not exaggerating. 1.5 or 5 hours is considered to be very fast for animations. You need to plan rendering if you work with 3d. Render stuff when you sleep for example.
Alright, thank you guys, it clarified it for me perfectly, I didn’t know that rendering could take that long though, but it makes good sense with the frames and such. Yea a good workaround could be too render while sleeping, or maybe even buying a computer only for 3D work, so it can run through daytime too.
I was also thinking about movies like Jurassic Park and the music video for Lil Dicky’s “Pillow Talk”, do the professionals behind these visuals just bite the bullet and wait for days or even months for their 3D animations to be done, or do they have supercomputers? XD
I’ve done a decently long video at a relatively brisk 5 - 10 minutes per frame. It took a half dozen computers running overnight for about a week.
The “External graphic card” mentioned might be up for debate - are you referring to a discrete GPU, such as a GTX 1070? Or an actual external device, through a Thunderbolt port, for example?
Time for rendering can depend on a number of factors. For example, if you have a low-grade GPU then you’re going to want to use the CPU for rendering; if you have a low-grade CPU then you’re going to want to use the GPU for rendering. If both are entry-level… stay away from grass, hair, particles, etc. and focus on simpler hard surface objects for the time being. In order to do the fancy stuff you’re going to need some decent hardware, maybe not super high-end but still substantial.
Don’t get hung up on doing everything all in one go, break it up into chunks that can run in a decent amount of time e.g. overnight.
Don’t use final/high-quality settings for test work, use limited image size/quality/FPS until you’re ready for generating the finished product.
But … there are many things that you can do to help improve things. Two of the biggest ones are (1) “breaking the problem down,” and (2) “deciding what, in each case, is good enough.” You will very quickly come to understand the value of compositing and the important role that it plays in the production process.
You’ll also discover what “OpenGL Preview” renders are all about – where you cut together the sequences of the show before final-rendering any of them.
As a simple example, I fairly recently was putting together a sequence to illustrate the functioning of an old-fashioned “twirling-balls” steam engine governor. Dealing just with one medium-close-up shot:
The entire sequence can be “previewed” very quickly before being broken down. (“Decide if you’re going to use the shot at all, and if so how much of it, now.”) (But also: “in preview, choices are relatively painless, and yet extremely accurate.”)
If the camera isn’t moving, everything that isn’t moving can be rendered as just one frame.
You only need to render one rotation, if the speed of the engine does not change for the duration of the shot.
You really don’t need shadows. No one is likely to notice that they’re not there.
When you do need shadows, sometimes you can best render them separately.
Remember that you can use Cycles, BI, and any other available rendering technology in parallel, if needs dictate. Yes, it takes more “setup,” but it can (maybe!) save a ridiculous amount of time.
What you do not want to do is to commit to some “big kahuna render” in which you’re basically rolling all your dice all at once and desperately hoping for the best," with your only alternative is to roll all those dice again. (Guess what you’ll wind up doing …) Every single shot requires careful planning, and very heavy initial use of previews – which are a very functional tool for “pre-visualization.”
Also: most of the time, you really don’t need to strive for “physical reality or accuracy,” even though Blender can usually achieve it. You need to consider carefully what you need for the shot to looklike, and to seek the most-efficient way to get “close enough” … by deadline.
Hi
Your advice really helps me getting a broader understanding of this topic. I think I will be turning to this thread a couple of times more to reread your answers and to put the understanding to use. And also I was referring to a discrete GPU, it was just something I heard in a video. That being said I really appreciate that you took your time to answer, because I did not know what to do about this, and know I have a way of working through it. Thanks to you all
