Now that the Eevee hype is over...Anyone use it for actual work?

Someone actually tried to make a comparison.

Though in the comments section people say that is not exactly the ideal test case in terms of lighting or materials, which makes sense, because using EEVEE you always know what you can do and can’t do, in terms of lighting conditions and picture fidelity.


Another thing to consider is the render time is different based on the computer. Eevee is faster than cycles, if I’m using my cpu. But if I use both of my GPUs with cycles, the result is actually faster. Unless I’m doing volumes. Eevee stinks at volumes, but they are so much faster then cycles. Am I doing something wrong? Is there a way to make the volumes actually look good in eevee?

Also remember that the fundamental algorithms used by Cycles vs. EEVEE are really completely different. They approach the problem, so to speak, from opposite directions. Both can give you excellent results – neither one is a panacea. “Volumes” are one way that you can very clearly see the difference between the algorithmic approaches.

1 Like

I just used it for this. These sorts of animations used to take so long to render.

It’s working well so far for me. Love the new instant motion blur.
It has limitations. But for a very large majority or renders and projects I think easily more than good enough. Especially for small teams or people working solo. It is looking like its getting to be a serious option for TV show animation production rendering now. And it seems to be improving at a fairly rapid rate.

I still miss the seemingly unlimited flexibility and freedom that there was when using the old Blender render in tandem with the compositor and render layers. But this functionality seems to gradually be returning with Eevee. Instant DOF now too.

I doubt if you can get good enough volumes in EEVEE, so you would have to do composite renders of both Cycles and EEVEE and merge the results with compositor.

This is how “Control” on PS5 looks. If you try to enhance the volumetric effect here, actually would be better done with post processing.
2021-03-17 17_39_29-ps5 volumetric - Google Search - Brave

Check this out:

Eevee is better than you (may) think

Just thought it was interesting.

Have you seen this?


Looks good.

That’s going to be great. It will be nice to be able to actually preview explosions and such in the viewport.

Indeed! I render at 960x540 and use topaz video enhance, upsizing the frames to 1920x1080 and the results are great. Might not be as good as native fullhd rendering, but reducing render times to one quarter is huge!

I use EEVEE for final work but use Cycles for any volumetric smoke/fire

This users Warhammer fan clip is EEVEE and looks good enough for my purposes


I guess the real question is, what is the ratio of additional setup time to the amount of time cycles took longer to render. I mean, eevee might render faster, but you have to worry about light probes, and tons of other settings to get it to look as good as cycles. With cycles, you can set it up easily and quickly.
Is it worth it to spend an extra two hours to have it render three hours quicker? Or can you get away with cycles, spending three hours more in render time, but saving two hours in setup time? (Completely arbitrary numbers, for sake of example)
Remember, you can do other productive tasks while rendering. However, setup time is way more valuable.

1 Like

I don’t think EEVEE requires that much more set up time than Cycles, I mean a bit per shot, but not an extra two hours.

Let’s say that Cycles can render at about 3 minutes per frame at final quality. That equals 12 hours of render time for just a 10 second sequence. Let’s say the same sequence can render in EEVEE at 30 seconds per frame, that’s 2 hours for the entire sequence.

Multiply that by the 3 minute short film linked above, and we’re not talking EEVEE saving a couple of hours, we’re talking EEVEE saving weeks!


Also … might I kindly suggest … “you really can’t meaningfully compare Cycles and EEVEE.” Even though the final objective is identical, their respective approaches to the problem are “fairly polar opposites.”

In my assessment, Cycles approaches the GPU as being “a very-convenient mathematical array coprocessor,” which it can exploit to do a series of calculations which the CPU can also do. Meanwhile, EEVEE engages the GPU to do the task for which it was originally designed. And, while this might sound like a contradiction, it really is not.


like you said EEVEE uses the GPU for what it “was” intended for back then, rasterization and all. Nowadays GPUs are getting more and more extra hardware (AI for denoising, RT cores for faster ray tracing, motion blur accelerators, etc…), all geared towards accelerating the ray tracing side (cycles & the like).

The speed gap is shrinking with each GPU generation, like it was said before on multiple forums, if you have a weak GPU, EEVEE looks more attractive, and you are happy to trade a few “realistic” cards for “speed” ones, but when you get one of those new shiny (future shiny) GPUs, the “speed” part becomes less and less of a problem and you are left with just “realistic” cards to play with.

Perhaps comparing complexity and memory size or server farms and individual workstations would be a useful reminder of different usage patterns?

By the way, cycles just got some new performance and speed updates:


I work almost exclusively in Eevee due to the fact that I don’t have a super powerful machine. I’ve made probably about 3 hours of animation in Eevee in the past year just of single scene renders and such. I’m planning on using it for an entire TV show that I am working on with some other artists and we will see if it holds up to the task.