Jump to content

octane vs redshift


Recommended Posts

Hello
I started studying c4d and now I'm going to study renderer for the first time. I'm thinking about which one to choose between octane and redshift. I am interested in artwork that composites live-action videos such as music videos, and I am interested in producing cinematic videos with realistic lighting. Can you recommend which render would be good to study?

Edited by Cerbera
removed formatting (see edit history)
Link to comment

Go for Redshift: more stable and uses  standard workflows that you can easily carry over to other render engines. On the downside, its biased nature implies many parameters that can feel overwhelming for beginners. 

Octane is often said to be better for quick photorealistic results, but it's much less stable and its workflow is more specific. 

 

 

 

 

Link to comment
1 hour ago, Adrien said:

Go for Redshift: more stable and uses  standard workflows that you can easily carry over to other render engines. On the downside, its biased nature implies many parameters that can feel overwhelming for beginners. 

Octane is often said to be better for quick photorealistic results, but it's much less stable and its workflow is more specific. 

 

 

 

 

Well in the latest release guys from RS seemed to improve that by adding different UX and UI, so it should be much easier  to get decent results from the get go.

Link to comment

if you are trying to archive realistic results (filmic) go for octane.

It is by far the easiest render engine to achive realistic render results,

If you need a versatile renderer for multi purpose go for redshift.

it is much harder to get realistic looking results, but it is a full fledged production render and can basically do what ever you like.

as both are available for subscription you can easily rent each for one month and decide yourselves.

 

 

Link to comment

So the main difference between the two is that Redshift is a "biased" renderer while Octane is "unbiased"

 

I for one prefer bias in my render engines.  Now, before we careen into the SJW discussions, let's first understand the difference between the two.

 

Biased renders allow you to set the number of ray bounces that can occur within the scene.  As such, there are many ways to do that as there are your camera rays (those coming from the camera), light rays, reflection rays, refraction rays, etc. etc. etc.  For each you can determine the number of rays being generated to render refractions, reflections, etc.  and the error threshold that the program uses to consolidate all that information.  For example, a camera ray will hit a highly reflective surface. From that point of contact you can determine how many "reflection rays" emanate from that point to capture the surroundings.  Now in reality the information flow on that ray is NOT from the surface to the surroundings but rather from the surroundings back to the surface and ultimately back into the camera.  That is important to understand as all that ray sampling is compared and Redshift looks to see that the values being calculated for each of the reflection rays when all compared together is below the error threshold you set before the reflection amount for that ray is calculated.  If you still see some issues with how the reflection  is being rendered (eg. too noisy) you increase the rays by a multiple of 2.  You play with each setting in the main menu and watch how the scene improves (shadows get crisper, noise is removed, etc).  Honestly, I have set ray values to 1024 and the render time actually improves while the scene looks better!!!

 

So what is the advantage of all this complexity?  Speed as you are directly controlling how and what the render engine needs to be concerned about.  For example, if there is a lot of motion blur, you may want to increase the camera rays (to remove noise from the blur) but drop the reflection rays as you really are not going to see it all that well anyway.

 

Now this sounds a little over-whelming but it is not (IMHO).  It also explains why people say it is hard to harness Redshift's capabilities.

 

Unbiased renders do not offer as much control.  They are unbiased because they use real world physics to set the number of ray bounces, etc.  The unbiased render engine will allow the number of bounces to continue until the energy in that photon is fully expended.  Not sure if that is how Octane works, but that is essentially the core difference between biased and unbiased render engines.

 

So the fact that unbiased render engines really simulate the physicality of light, is why they say you can get great results right out of the box.  The downside of unbiased render engines is that you do lose some control and the scene may not render as fast as what you can get from an optimized biased render engine.  Going back to my original example, if you don't care about reflection quality in a fast moving scene with motion blur and you may find yourself with less options to manage that in Octane than you would in Redshift.  Again, I am not an Octane user but I just wanted to make the point between how you can optimize render times in biased vs unbiased render engines.

 

Honestly, I think the best way to approach this is to ask yourself if you will be doing stills or animations.  For stills, you may want to get great results with minimal fuss and the extra render time to get that is not a big deal.  

 

If you are doing animations, optimizing the render time per frame with a biased render engine could be a huge benefit.   If you can cut render time by 30 seconds per frame on a 10 second  (300 frame) animation - that is 2.5 hours you just saved.

 

Hopefully that makes sense.

 

Dave

Sorry...but I simply do not have enough faith to be an atheist.

Link to comment
48 minutes ago, 3D-Pangel said:

So the main difference between the two is that Redshift is a "biased" renderer while Octane is "unbiased"

 

I for one prefer bias in my render engines.  Now, before we careen into the SJW discussions, let's first understand the difference between the two.

 

Biased renders allow you to set the number of ray bounces that can occur within the scene.  As such, there are many ways to do that as there are your camera rays (those coming from the camera), light rays, reflection rays, refraction rays, etc. etc. etc.  For each you can determine the number of rays being generated to render refractions, reflections, etc.  and the error threshold that the program uses to consolidate all that information.  For example, a camera ray will hit a highly reflective surface. From that point of contact you can determine how many "reflection rays" emanate from that point to capture the surroundings.  Now in reality the information flow on that ray is NOT from the surface to the surroundings but rather from the surroundings back to the surface and ultimately back into the camera.  That is important to understand as all that ray sampling is compared and Redshift looks to see that the values being calculated for each of the reflection rays when all compared together is below the error threshold you set before the reflection amount for that ray is calculated.  If you still see some issues with how the reflection  is being rendered (eg. too noisy) you increase the rays by a multiple of 2.  You play with each setting in the main menu and watch how the scene improves (shadows get crisper, noise is removed, etc).  Honestly, I have set ray values to 1024 and the render time actually improves while the scene looks better!!!

 

So what is the advantage of all this complexity?  Speed as you are directly controlling how and what the render engine needs to be concerned about.  For example, if there is a lot of motion blur, you may want to increase the camera rays (to remove noise from the blur) but drop the reflection rays as you really are not going to see it all that well anyway.

 

Now this sounds a little over-whelming but it is not (IMHO).  It also explains why people say it is hard to harness Redshift's capabilities.

 

Unbiased renders do not offer as much control.  They are unbiased because they use real world physics to set the number of ray bounces, etc.  The unbiased render engine will allow the number of bounces to continue until the energy in that photon is fully expended.  Not sure if that is how Octane works, but that is essentially the core difference between biased and unbiased render engines.

 

So the fact that unbiased render engines really simulate the physicality of light, is why they say you can get great results right out of the box.  The downside of unbiased render engines is that you do lose some control and the scene may not render as fast as what you can get from an optimized biased render engine.  Going back to my original example, if you don't care about reflection quality in a fast moving scene with motion blur and you may find yourself with less options to manage that in Octane than you would in Redshift.  Again, I am not an Octane user but I just wanted to make the point between how you can optimize render times in biased vs unbiased render engines.

 

Honestly, I think the best way to approach this is to ask yourself if you will be doing stills or animations.  For stills, you may want to get great results with minimal fuss and the extra render time to get that is not a big deal.  

 

If you are doing animations, optimizing the render time per frame with a biased render engine could be a huge benefit.   If you can cut render time by 30 seconds per frame on a 10 second  (300 frame) animation - that is 2.5 hours you just saved.

 

Hopefully that makes sense.

 

Dave

I would have been 100% with you on rendertimes even 3 years ago. But with the new grafics cards and speed increases in octane it is really no thing anymore. You can of cource kill every sceene by doing unnecesarry things (push all boxes) but octane on modern hardware is fast enough for animation work without a doupt. You can still render faster in redshift, but then you will have to make the calculation how much money spare time is worth. optimizing a project for a day, or buy a additional grafics card.

Link to comment
27 minutes ago, Jops said:

I would have been 100% with you on rendertimes even 3 years ago. But with the new grafics cards and speed increases in octane it is really no thing anymore. You can of cource kill every sceene by doing unnecesarry things (push all boxes) but octane on modern hardware is fast enough for animation work without a doupt. You can still render faster in redshift, but then you will have to make the calculation how much money spare time is worth. optimizing a project for a day, or buy a additional grafics card.

Fair enough.  Again, as pointed out in the post, I am not an Octane user and my examples were more focused on the upsides and downsides of biased vs. unbiased renders more so than the differences between Octane and Redshfit actual render times.

 

Now, I personally have found that optimization is not a day long event as you stated.  Rather, at most one hour for average scenes.  You could optimize Redshift for each texture which can be a bit of overkill for the hobbyists like myself (though required for SSS).  I tend to work at the scene level which will  not produce 100% optimization that a professional would do who is trying to trick out their production pipeline for a big job, but it is good enough for me.

 

Now what I do like about Octane is that they are partnering with plugin developers like World Creator and Embergen.  Fluids and environments are definitely two areas that can tax a render engine.  Now Redshift actually promotes integration with Autodesk plugins more so than C4D plugins - which is kind of odd after being owned two years by Maxon.  While there are some settings for importing VDB files from X-Particles (which is appreciated) more could be done on the integrations side.

 

Dave

Sorry...but I simply do not have enough faith to be an atheist.

Link to comment
6 hours ago, MJV said:

It's no contest imo. Octane blows Redshift to smithereens.

 

Yeah I can't agree with that. Even though I too PREFER Octane over RS, I wouldn't say it blows RS out of the water by any means. For a start it is considerably less reliable and much more crash-happy than RS. And slower to render, given that RS is about the fastest renderer there is, and also among the most stable.

 

For me Octane wins because it gets photo-real results rather more easily than RS can and I like its nodes more !

 

CBR

 

 

 

Link to comment

I'd say Octane is better for stills & concept work, Redshift for Animation. As for stability, it's all anecdotal, mostly. I'm sure that they both become more & less stable over time as the codebase changes & is revised & depending on which bits of the software you're using.

Link to comment
14 hours ago, Mike A said:

 

In what sense?

In the sense of what the OP is asking for. Having worked with the highest end video and optics my entire professional career as a DP, in my mind there is simply no comparison in terms of photorealism, and that's what the OP here is asking for. When I first saw the opening credits for Westworld and then tried Octane for myself, I truly thought I had died and gone to heaven. An unbiased renderer physically accurately resolves light and all the physical behaviors of light by tracing the paths of reflection and refraction to final termination. Octane was not the first unbiased renderer on the market but was the first fast enough to make it a viable solution for animaion. Unlike an unbiased renderer, a biased renderer like Redshift composites various processes together to arrive at an approximation of physical reality, all intended to make the solution faster, more versatile and more customizable, and for certain jobs is still a worthy contender, especially for architectural work and the like and especially in environments where the light source or sources are very relatively tiny, or for artistic or non photorealistic renders where control is more essential than realism, and I believe Redshift is amoung the best solutions currently available for that sort of thing. But when the question is about photorealism, Octane is the real deal. On a side note, it's always been rock solid for me.

Link to comment
×
×
  • Create New...

Copyright Core 4D © 2023 Powered by Invision Community