Chaos Corona Forum
General Category => General CG Discussion => Topic started by: lupaz on 2017-11-05, 13:48:50
-
Hey guys,
Redshift is coming across often for some reason that I cannot figure out.
Even if it’s the fastest render engine, the images I’ve seen are far from realistic looking. They look like made with vray 1.
Does anyone know what’s the buzz around it for?
-
That's rather under-selling it.
Judging by what pictures ? The whole Blizzard Cinematics team is using it to produce absolutely amazing...Blizzard level quality.
I've played with Redshift about year ago and some of my observances:
-It's actually fast GPU renderer. That cannot be said about any other GPU renderer on the market, Redshift really is ahead, massively. But it has drawbacks of course, that speed is either at expense of GI (using general brute force), or quality/precision because you have to use (or rather, have the option to use) Irradiance caching for interiors. I presume some of those IR renders can resemble old Vray renders because it's the same underlying tech.
-It had great PBR shader from second version, and also supported Arnold's AlShader at same time. That's two industry level PBR shaders at same time with all the nice things to boot (correct fresnel, multiple BRDFs like GGX and Oren-Nayar)
-It was the first (and still the only?) GPU renderer capable of out-of-core rendering, cycling the limited GPU's memory buffers. This is massive limitation for big productions.
It's basically GPU speeded up Arnold for the masses. More than well deserving its praise imho :- ).
Now there are renderers running almost purely on hype and strangely religious cargo-culting, but Redshift is not among them.
-
Thank you Juraj.
My comment about vray 1 was as a joke, implying that the quality of the image feels a step backwards compared to Corona.
I understand what you say. Big productions benefit from those advantages. But for the individual artist, other than speed... I would assume they'd choose realism over technical benefits. In this sense another advantage has to be ease of use I imagine.
It also seems to be popular among game artists and NPR rendering artists. Not so much architectural, no?
-
Imho the reason why Redshift is basically non-existent in archviz is simply because they haven't tried to get into that market yet. Redshift currently is a vfx renderer, and that's very apparent. I have spent quite some time just last week to test it out and I'm really impressed so far. However it's NOT a corona. Not in terms of ease-of-use and not in terms of archviz-oriented features. I don't share the opinion about GI (you can use the same methods as in vray/corona, it appears to be pretty solid actually, for interiors as much as everything else) but there are some features a lot of corona users will have a hard time to get used to - like defaults that are optimized for speed, not realism. Secondary rays for example are aggressively clamped by default (but you can set it higher or disable it of course). I also don't think the "quality" is a step backwards. Afaik there is no technical reason why redshift wouldn't be able to achieve as realistic renderings as corona (apart from the yet missing raytraced SSS, the point based one they have is pretty nice however).
-
When I used brute-force on Interiors, it was slower and way noisier than Corona, but it was year ago, perhaps it did improve massively since, which I could see happening.
-
I was actually quite surprised. I'm not rendering many interiors, so you'll have way better scenes to test with than I do, but the ones I tried were quite good. However it's relatively easy to oversample with redshift, so you have to make sure you setup the GI samples and the universal sampling settings correctly. A "simple" cathedral test I did a while ago: https://www.facebook.com/martin.geupel/posts/10207971459843258 (https://www.facebook.com/martin.geupel/posts/10207971459843258)
-
Ok, your test fared much better ! That's a very good scene to test.
I did play a lot about the settings, as I am confident I understood the whole sampling engine with old 2.4+ Vray rather good as well. Perhaps it has to do that my 20c/40th workstation was simply ahead of single Maxwell Titan-X.
But if that's the result a single 1060 can get..I can only imagine how happy are the people with 6x1080Ti, something not too problematic to built if it's your primary workstation and node at same time.
Actually makes me bit jealous too..
-
I'd love to have it as my shading mode directly on the viewport. It seems we're getting there in terms of speed.
-
When I used brute-force on Interiors, it was slower and way noisier than Corona, but it was year ago, perhaps it did improve massively since, which I could see happening.
I was just trying the demo and yes, I feel it is like you say.
Corona seems to be faster on interiors, especially having the denoiser.
The real time experience in Corona seems to cover more features. I have to refresh the VFB in redshift somtimes.
But on exteriors it's blazing fast. It'd be nice to have for exterior massive animations.
The best of all is the fact that is GPU based. My system runs so silent with it. But then there's Octane and Fstorm...
I hope Corona GPU is in the making.
-
When I used brute-force on Interiors, it was slower and way noisier than Corona, but it was year ago, perhaps it did improve massively since, which I could see happening.
Did you use BF / BF or did you do some extra "bias" magic with the second solver?
-
You know what, I feel like testing it again :- ).
-
You know what, I feel like testing it again :- ).
Zhiyao Chen provided a cool and practical scene for testing on his blog -> http://zchen.ca/gpu/
Don't forget to report in with the findings :)
-
To be honest, if you had €5,000 to spend on graphics cards it might be work using as a a production tool.
But for large Arch jobs all the GPU setups are not usable.
Main Model, trees, railclone, cars, HD maps...........
-
To be honest, if you had €5,000 to spend on graphics cards it might be work using as a a production tool.
But for large Arch jobs all the GPU setups are not usable.
Main Model, trees, railclone, cars, HD maps...........
Not sure I understand. Do you mean that memory wouldn't be enough?
-
To be honest, if you had €5,000 to spend on graphics cards it might be work using as a a production tool.
But for large Arch jobs all the GPU setups are not usable.
Main Model, trees, railclone, cars, HD maps...........
Not sure I understand. Do you mean that memory wouldn't be enough?
Yes, one of the main reason why I switched from GPU to CPU (Corona) renderer :)
can' t go back to GPU. I would rather do a primary/secondary rendering choice and use Corona as my primary one for complexe or big projects the VRAM limitation is nothing compared to GPU.
Right now, the main problem is : you can buy as much 1080ti (11gb VRAM I think let's say 4 of them, so 44gb you will say? Well, not even close, it will be using 11gb. And some renderers are even worst in term of VRAM limitation : let's say you have a GTX 970 4gb and a 1080ti 11gb, the renderer will use the lowest VRAM (4gb) and not use the 11gb.
you switched from GPU renderer to CPU because of vram limitation, I can see this with octane or vray RT, but Redshift is a different animal. It can handle insanely large scenes compared to other GPU renderers.
-
afaik, redshift will use the system's memory when the GPU runs out of memory, but will decrease the render speed.
https://www.redshift3d.com/support/faq
When Redshift uses multiple GPUs, is their memory combined?
Unfortunately, no. Say you have an 8GB GPU and a 12GB GPU installed on your computer. The total available memory will not be 20GB, i.e. the 8GB GPU will not be able to use the 12GB GPU's memory. This is a limitation of current GPU technology and not related to Redshift in particular. We, therefore, recommend users combine videocards that are fairly "equal" in terms of memory capacity.
Having said that, Redshift supports "out of core" rendering which helps with the memory usage of videocards that don't have enough VRAM (see below). This means that, in contrast with other GPU renderers, the largest possible scene you'll be able to render in the above scenario won't be limited by the system's weakest GPU.
What is "out of core" rendering?
Redshift has the capability of "out of core" rendering which means that if a GPU runs out of memory (because of too many polygons or textures in the scene), it will use the system's memory instead. In some situations this can come at a performance cost so we typically recommend using GPUs with as much VRAM as you can afford in order to minimize the performance impact. Certain types of data (like textures) actually work very well with out-of-core rendering. This means that even if your scene uses tens of 4K or 8K textures (i.e. several GB worth of data), you can still expect great rendering performance!
-
Hi NEO-N,
Not sure what GPU renderer you used was but Redshift is different from all the other GPU renderers. You said the GPU renderer you used (not Redshift) you was limited to 15M polys, that's horrible, or your graphics card was junk. You can fit 15M polys on a card with 1.5gb VRam easy, think that's a $80 video card. I can tell you from one of my projects that was around 60M polys we rendered just fine and unbelievably fast on a $300 GTX 980ti with 6GB Vram, no problem. That was pure polys before any instancing mind you. You say Corona is unbiased, you mean Pathtracing + Pathtracing, doesn't most use Pathtracing + HD Cache cause it looks pretty much identical but much faster to render? I mean in Redshift you can also do Brute Force + Brute Force if you want an (un)biased rendering, but BF + IPC is like Corona PT + HDC, and output is identical to Corona. You even have more options In Redshift for Biased rendering, but you don't really need it since its so fast. A lot of people try Redshift and don't know what there doing, I know when I first tried I was using it wrong, I thought the progressive renderer was the final Renderer and even though it was still fast and good, I found out after bucketing rendering is where all the magic happens. We recently did a animation and I tried Corona 1.7 for one of the rooms to see the outcome, did a great job, but you couldn't tell the difference from the Redshift render. I think there was one issue with the shader or something, not sure if it was the shader or core, but from using shaders from various renderers I can say by far Redshift Material is the most advanced material shader I have used.
So I see a lot worried about vram and scene size, I was too at first until I realized how huge a scene Redshift can actually render with no issues. Other GPU renderers I definitely had issues with even much smaller scenes. Soon we will also have NVlink which will share memory with cards, then Vram issues will be gone for even the most extreme cases.. For now those are only with the highest Teslas I think.
-
I presume some of those IR renders can resemble old Vray renders because it's the same underlying tech.
Yup. It looks exactly like vray. The irradiance cache is a copy.
-
there are so much suffer to live in these days when there are so many gpu renderers out here
-
You know what, I feel like testing it again :- ).
Hi Juraj,
I was wondering if you tested it again. I'm interested in what you have to say about it.
-
I didn't yet...but Deadclown did. And his results are impressive and intriguing.
-
Thanks Juraj :)
Since this is the corona forum I think it's not right to list the advantages of a competitor here. Redshift also has quite a few drawbacks but half of what has been said is just not correct the way it was said.
Long story short: redshift is not a one-click renderer (as corona is) and requires time to read the documentation and learn the details. In terms of quality I still see no reason why you shouldn't be able to produce as high quality renderings as in corona - but I definitely think that a lot of corona users won't have fun with redshift if they don't like vray.
I suggest to have a look in this Facebook group https://www.facebook.com/groups/RedshiftRender for a better "overview" of what is being done with Redshift these days (instead of their gallery). One example for the archviz business (and for completely nonsense rendersettings unfortunately): https://www.facebook.com/groups/RedshiftRender/permalink/376257112823505/
-
Pretty sure I've seen an Ondra smiley face with regards to implementing Nvidia's denoising....
The redshift teaser looks like we will be in for a treat with it.
-
So it calculates denoising on the fly? Interesting. Also curious to know if it's limited by available amount of GPU RAM?
-
My first impressions are that it does worse job than Corona's denoising, but of course it is much faster, close to real-time. We will have to wait a bit to see it in action, and then have a better understanding of it.
So far it seems that there are some use cases where it makes perfect sense to use it (IR?), and in other cases it may be useless.
Also I bet it comes with some Nvidia- or general GPU-related limitations. ;)
-
I'm guessing it's running on a ninja card(s) in the video but interesting nonetheless for the fact that RS is maxing the card(s) out anyway when rendering whereas Corona (presumably) doesn't touch the graphics card at all at the moment so I'd like to think it might be even quicker on Corona?
-
It would be insane if Corona could implement it so those useless GPU-s are moving a little bit :D ..I bet it would be lightning fast since CPU is doing all other work :)
-
- VRay is getting it with v4... see it @ Vlado's YT channel (http://see it @ Vlado's YT channel)
- Clarisse also (which is same as Corona, CPU engine)
- On Iray, denoiser uses ~800MB VRAM alone. (source (https://forum.nvidia-arc.com/showthread.php?16422-AI-Denoiser-for-3ds-MAX&p=67044#post67044))
The way the denoiser is implemented is, that it uses a soft-limit of 800 MB of video memory for denoising. I am saying soft-limit because in certain cases the memory consumption may go beyond that limit, but in general it is pretty solid.
Up to 800 MB might seem like a lot, on the other hand you get a big quality increase for that.
Best Regards,
Rüdiger
Yes, CUDA tech only. Working in unison with CPU. Throw cryptomatte into the equation & you get to play as Vlado does above - almost Real time ;) ... am already missing moments when i had real time to contemplate, think & make decisions on the work, it's design... but it's fine. Now more invested in conceptualization & the environment is quieter, more stable (less can go wrong in couple of minutes then in couple of hours) :D
-
The speed is amazing, but when he zoomed VFB and switched between raw and denoised images... yikes, that is exactly the type of denoising i hate - smooth, featurless, plastic fantastic. I can get the same with Corona denoising, if rendered only 2 passes :]
-
The speed is amazing, but when he zoomed VFB and switched between raw and denoised images... yikes, that is exactly the type of denoising i hate - smooth, featurless, plastic fantastic. I can get the same with Corona denoising, if rendered only 2 passes :]
Yeah, but it will probably take a bit more time to denoise. ;)
Also the quality would be probably better after more passes.
-
yes, it can be used progressively
-
Real time denoising would be nice for IR, especially if it comes from the GPU.
But to be honest the only thing I envy from GPU render engines is that you don’t hear the jet engine fans of CPUs. It sounds a bit archaic. And the scalability on a single workstation of course.
Any other feature is like meh.
-
Real time denoising would be nice for IR, especially if it comes from the GPU.
But to be honest the only thing I envy from GPU render engines is that you don’t hear the jet engine fans of CPUs. It sounds a bit archaic. And the scalability on a single workstation of course.
Any other feature is like meh.
Heh, afaik its not a treat to hear 4x 300W GPUs working at a 100% either :) Unless you have liquid cooling that is but then again the same can be done to CPUs.
Totally agree about scalability though. I think thats a super interesting aspect of GPU rendering. 3x GPUs not fast enough in IR? Plug in 3 more! :)
-
Yeah, that was bit odd comment :- ). I would suspect either poor case, CPU cooler, or setup.
Personally, I find CPU tower coolers to be extremely silent, they are for most part massive 140mm fans on heatsink that don't need to spin a lot (<1000 RPM to maintain even high overclock).
My 2x Noctua NH-U14 spin only up to 900 RPM to cool my dual-xeon (2x140W), generating roughly <= 26 dB. It's inaudible almost, I can only hear them when everything else is shut down at night.
Oppositely, if you want to stack a lot of GPUs next to each other, it's preferable to use blower-style reference models, and those generate the most audible noise I am aware of.
This means the Founder's edition of 1080Ti easily reaches 42 dB because it spins up to 4000 RPM. That's jet engine :- )
Only way to avoid it is to use custom water-loop, but you need massive pump to go through multiple GPUs and those pumps spin 2000+ RPM very easily, making it again, rather audible. Such setups are never <30 dB inside massive case.
But of course, scalability is amazing. That's the best feature of GPUs bar none.
-
...
Oppositely, if you want to stack a lot of GPUs next to each other, it's preferable to use blower-style reference models, and those generate the most audible noise I am aware of.
This means the Founder's edition of 1080Ti easily reaches 42 dB because it spins up to 4000 RPM. That's jet engine :- )
Only way to avoid it is to use custom water-loop, but you need massive pump to go through multiple GPUs and those pumps spin 2000+ RPM very easily, making it again, rather audible. Such setups are never <30 dB inside massive case.
...
I hate to nitpick your posts but from what I'm reading up on the subject I'd suggest a hybrid GPU cooler as a good in between option . It usually has a single / two fan design (one on the cards the other on the radiator) and the cooling performance is way better than with the rear exhaust cards when you have 4x GPUs side by side because you don't rely on fresh air intake from inside the case where all your GPUs are.
I haven't tried it but I'd say its a good in between option because you now have fairly large fan spinning plus it doesn't need to rev up all the time due to the radiators being outside the GPU area.
-
...
Oppositely, if you want to stack a lot of GPUs next to each other, it's preferable to use blower-style reference models, and those generate the most audible noise I am aware of.
This means the Founder's edition of 1080Ti easily reaches 42 dB because it spins up to 4000 RPM. That's jet engine :- )
Only way to avoid it is to use custom water-loop, but you need massive pump to go through multiple GPUs and those pumps spin 2000+ RPM very easily, making it again, rather audible. Such setups are never <30 dB inside massive case.
...
I hate to nitpick your posts but from what I'm reading up on the subject I'd suggest a hybrid GPU cooler as a good in between option . It usually has a single / two fan design (one on the cards the other on the radiator) and the cooling performance is way better than with the rear exhaust cards when you have 4x GPUs side by side because you don't rely on fresh air intake from inside the case where all your GPUs are.
I haven't tried it but I'd say its a good in between option because you now have fairly large fan spinning plus it doesn't need to rev up all the time due to the radiators being outside the GPU area.
And where would you conveniently place the fans for these hybrid GPUs ? 4x Hybrid GPUs on top are 4x 120/140mm fans to place somewhere in the case.
The on-cards fans will still obstruct themselves in the slots in airflow.
And last, that might be the most audible solution imho, as you'll have 4 crappy micropumps. Not fan of hybrids.
-
To me the benefits seem to come from the fact that you potentially have 4x 120mm fans running at low speeds. In Fractal Design XL R2 I can find a few places where I could put them without modding the case :)
The on-card fans are obstructed but the same is true for the rear exhaust ones. I could be wrong but afaik the VRM chips benefit the most from that fan and it isn't that big of a deal because you are facing the same scenario (if not worse due to higher temps of the GPU) with blower cards. Granted, it probably really helps if you have a decent flow in your case.
That said VRMs do tend to get hot.
So to me thats 4x 120mm fans at lower speeds versus 4x something thats considerably smaller at high rpm (its fairly well documented that 4x reference blower equipped GPUs throttle fairly often which means its probably as loud as you've pointed out).
Since GPUs / CPUs get really inefficient when overclocked you could _potentially_ further silent the hybrids by restoring them to near reference clocks.
But I do agree that a full out water solution is more silent, for sure :) Have you had a bad experience with a hybrid GPU? If so, which one did you use?
-
funny thing, between 2007 and 2017, the best comment for derailing a thread on any technical forum stayed the same - computer noise :D
-
But I do agree that a full out water solution is more silent, for sure :) Have you had a bad experience with a hybrid GPU? If so, which one did you use?
None. I consider them waste of money. 4x hybrids pay for custom water loop by themselves.
-
@Ondra, Which is totally funny considering I personally don't really care that much about noise yet here I am debating it :)
@Juraj, Well I think thats rather bold? The average "cheap" 1080ti comes in at around 740 in .de (I'm from Slovenia tho but lets not overcomplicate). If you put an EKWB on it just the GPU cover alone costs 110 eur which brings it to around 850 EUR per card. You gotta add costs like the radiator + the other screws and bolts which adds up what, another 100-200-300? :)
An EVGA 1080ti (factory OC) hybrid is cca 870 EUR.
Dunno, just seems like if you don't want the hassle / paying extra for the full loop the Hybrid way seems like a good option to me. Hopefully someone else could chime in to set us both wrong :)
As for Redshift itself, right now the tonemapping options are pretty limited. I heard they are tackling that one as well but its a ++ feature that I like about Corona :) Gotta throw a kudos to the devs :P
-
I feel like you are arguing just for the sake of bringing forward niche, unorthodox solution that's rarely if ever used as some real common alternative.
If you build workstation with 4 gpus, that's solid investment were additional money towards loop doesn't make much of a difference.
But trying to fit 4 hosed fans would either compromise airflow of regular tower cases, or require water-cooling towers that have many fan openings available and are foremost made for.... water loop.
I googled such builds to see how popular it is and it look quite ridiculous. It could be viable perhaps for up to dual-GPU setup, but after that I am quite sceptical.
(https://www.fuchsundvogel.de/perch/resources/dsc2374-1-w640.jpg)
-
i see... it's mainly because of maya :p
-
i see... it's mainly because of maya :p
Agree!
I came to that conclusion too.
-
@Juraj
Nah, not arguing just for the sake of arguing man... Debating :) Its a solution I know that a few really good artists use and is something I am considering for myself too. I have other reasons why I think a full water solution wouldn't work for me personally - but thats a different story. Like I said tho, performance wise its the preferred solution but so is a 2x2699 build. Not everybody has one obviously.
I think its especially fitting for those that go the middle route, something like a hybrid 1070 x 4 and are mindful of their budgets.
-
Juraj, Nkilar,
May I ask you a stupid question?
With all the bitcoin mania going on, aren't there self contained GPUs that can be plunged in externally to a workstation? That way you wouldn't have a overheating problem due to lack of space...
-
@lupaz
Welcome to the party!
I didn't see self-contained GPUs except for those Thunderbolt boxes that are mainly for Macs / laptops. I presume you mean those?
Those are all fun and great but the box alone costs 400$ or at least I did a year ago so for me its a little too expensive :) I think its a life safer for all those people running Octane in C4D and are using Macs or laptops.
Without water cooling I prefer the open workbench for performance. Its basically a computer on a desk without the case. Gotta get rid of your cats, dogs and possibly children I suppose, don't want no short circuitry :P I'm not using that though because I like cats, dogs and children :)
-
I used both in 3dsmax, you can hardly compare these two products without understanding the fields where they operate. Corona render is a standard of architectural visualization!
Fastest Cpu render plugin and reliable on huge scenes, can easily setup as mentioned before (one click render), tons of premade materials and tutorials for Archviz.
One real competitor is - Fstorm (which is also oriented in Archviz-3Dsmax).
Redshift in another hand is a best choice for vfx, motion graphics, films. Fastest and reliable Gpu render plugin with great functionality, you can tweak almost everything what you need in almost every software.
But doesn't have enough premade materials, tutorials, pages or communities for 3dsmax.. "rocket science setup" in the eyes of Corona user.
Render legion should work in the direction of Gpu render, otherwise they will start to loose step by step to someone like Lumion.. One click render feature heh? How about one click project? :)
Instead of developing plugin for Cinema4d, you guys should go deeper into Sketchup.. adding more functionality for vfx only will harm your product. And let's be honest, its really hard to beat the titan like Redshift(GPU) in Cinema4d or Houdini.
Wish you will stay in Archviz field :)
-
I used both in 3dsmax, you can hardly compare these two products without understanding the fields where they operate. Corona render is a standard of architectural visualization!
Fastest Cpu render plugin and reliable on huge scenes, can easily setup as mentioned before (one click render), tons of premade materials and tutorials for Archviz.
One real competitor is - Fstorm (which is also oriented in Archviz-3Dsmax).
Redshift in another hand is a best choice for vfx, motion graphics, films. Fastest and reliable Gpu render plugin with great functionality, you can tweak almost everything what you need in almost every software.
But doesn't have enough premade materials, tutorials, pages or communities for 3dsmax.. "rocket science setup" in the eyes of Corona user.
Render legion should work in the direction of Gpu render, otherwise they will start to loose step by step to someone like Lumion.. One click render feature heh? How about one click project? :)
Instead of developing plugin for Cinema4d, you guys should go deeper into Sketchup.. adding more functionality for vfx only will harm your product. And let's be honest, its really hard to beat the titan like Redshift(GPU) in Cinema4d or Houdini.
Wish you will stay in Archviz field :)
Same thoughts here...Gpu plz!
-
I'd love to see some current hardware price / performance metrics.
A new Ryzen 3900x compared to a 500€ equivalent graphics card, like the RTX 2070.
My feeling being, that Corona and CPU rendering in general still holds the upper card or equals GPU rendering in performance per dollar.
-
I'd love to see some current hardware price / performance metrics.
A new Ryzen 3900x compared to a 500€ equivalent graphics card, like the RTX 2070.
My feeling being, that Corona and CPU rendering in general still holds the upper card or equals GPU rendering in performance per dollar.
I have no scientific measures to offer but Vlado (V-Ray) posted some benchmarks with V-Ray on some of the Evermotion scenes. In general without RTX a 1x 2080ti was about as fast as a 2990wx (Running the CPU renderer).
From my experience for "simpler" scenes the GPUs are just plain faster (perf per dollar wise!!), no way around it imho. For heavier scenes and tougher interiors I think the playing field evens out - Again, that is just my impression. Then again, I'm using Octane which is PT / PT while Corona is PT / UHD for interiors. V-Ray offers BF / LC on the GPU and Redshift can be as biased you want it to be. So I have no idea how that scales.
Generally speaking though, with Octane at least, if the scene is brutal for pathtracers then the field will equalize fairly quickly since Corona has the UHD cache going. Interiors aren't much of a problem imho but an interior with 100x semi glossy spheres and a crap ton of SSS... Well... Thats tough.
RTX appears to give anywhere from 1.2x to 2.2x of a speed boost in interiors given some of the example scenes I've seen. I haven't tested it out though. For exterior stuff it however apparently speeds things up to like 7x. Again, haven't tested it.
So yeah it is tricky to compare directly. Not only you got RTX and no RTX, you also have BF / PT / Caching to consider.
-
I keep seeing people in the Fstorm Facebook page with hardware related problems (GPU). So it may also be fair to say that CPU is more stable (?).