Results 1 to 15 of 15

Thread: Your preferred workflow when determining render settings

  1. #1

    Your preferred workflow when determining render settings

    Browsing through the forums, it seems like everyone has varying opinions on this.
    I'm curious what process everyone uses when tweaking camera samples, reflection/refraction/sss samples, light samples, AA, etc...

    I personally start with the LW default settings (except for reflect/refract/sss samples, which I always bump to 3 to start) and then work through the areas I see noise using buffers, and bump up the appropriate sample settings 1 at a time. Then I go back and adjust camera samples (my understanding is that this setting is a multiplier of the other sample settings). I often loosely follow the "Removing Noise Workflow" in the LW docs https://docs.lightwave3d.com/lw2019/...noise-workflow

    I'm wondering if there is a technically correct order of operations when setting these things. Do you guys start with camera samples instead and work through other settings after? Do you always start with your own preset render settings? Do you adjust light sample settings right when you create the light or wait to adjust those until you're ready to render?

    Also, I rarely tweak any of the Limits (diffuse limit, recursion limits, ray precision) or the minimum camera samples for a scene. When/how do you decide to adjust those in your workflow?

    I'd love to see some screenshots of what everyone's final render settings look like for a 'typical' render (Yes, I know these settings are all very situational, just looking for examples) for stills or animation.

  2. #2
    Curmudgeon in Training Ma3rk's Avatar
    Join Date
    Mar 2003
    Location
    Near Beaverton, OR
    Posts
    1,433
    Don't know if there really is a "One Setting for All" as it depends on a number of factors.

    The Wiki page isn't a bad starting point. But in lieu of purchasing his series (which you might wish to do eventually anyway), you will benefit greatly by taking the time to view RebelHill's youtube vids. Definitely not hand-holding & not the entire series, but seeing his workflow & reasoning will be enlightening.

    https://www.youtube.com/channel/UCkf...Aj-az1iIeL0wdg
    "Never be a cat in a cartoon. Never." Chief Wiggum

  3. #3
    I agree, there's definitely no "one setting fits all". But i'd be curious to see the variances in individual render settings all the same.

    I've gleaned quite a bit from Rebel Hill's videos on the subject already, I might dig into some of the render-specific vids again.

    But I wasn't really looking for tutorials on the subject per se, I'm just interested to see how different people approach the process and the various workflows people use when adjusting render settings.
    Last edited by SilverFalcon; 10-09-2019 at 01:03 PM.

  4. #4
    Goes bump in the night RebelHill's Avatar
    Join Date
    Nov 2003
    Location
    jersey
    Posts
    5,770
    Quote Originally Posted by SilverFalcon View Post
    I'm wondering if there is a technically correct order of operations when setting these things...

    I personally start with the LW default settings (except for reflect/refract/sss samples, which I always bump to 3 to start) and then work through the areas I see noise using buffers, and bump up the appropriate sample settings 1 at a time. Then I go back and adjust camera samples...

    I'd love to see some screenshots of what everyone's final render settings look like for a 'typical' render (Yes, I know these settings are all very situational, just looking for examples) for stills or animation.
    There is such a way, and what you're doing is not it... Start with the camera, then lights, then the rest. You can, ofc, do it any way round, but doing it the correct way will result in the lowest achievable render time for a given level of quality. All other approaches will do nothing more than waste time rendering.

    And no, there is no typical render, and the settings arent 'very' situational, they're completely so... There is no basic 'standard' of settings that make a good starting point... it's nothing but a waste of time to think there is.

    The workflow IS the standard, you follow it exactly the same way, for every scene, every situation, and it works every time.
    LSR Surface and Rendering Tuts.
    RHiggit Rigging and Animation Tools
    RHA Animation Tutorials
    RHR Rigging Tutorials
    RHN Nodal Tutorials
    YT Vids Tuts for all

  5. #5
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    15,100
    Generally I adjust camera samples to something quite low, and since I often use volumetrics, the next thing I do is to raise light volumetric samples, which I feel is too low in itīs default state, it helps to acheive faster vpr drafts..but only so far as to get a preview of the volumetrics.. it will still look crappy if not using a higher amount.

  6. #6
    Registered User
    Join Date
    Aug 2016
    Location
    a place
    Posts
    1,843
    there are render presets, right?

    should get a set going for starting points. Maybe a set of example scenes would be better?

  7. #7
    Quote Originally Posted by gar26lw View Post
    there are render presets, right?
    Not that I am aware of. And the sample scenes in the content aren't really optimized for rendering from what I've seen. I've made my own start scenes for product shots, but they they wouldn't be good for much else. For interiors I start from scratch on each job and learn something each time. I also do quite a bit of experimentation as well, but if you really want to learn some good practices to get you off on the right foot invest in RebelHill's video series. It was money well spent for me. I'm experimenting with super hi-res rendering with no AA samples and the results are promising. Fair rendering times, some jaggy's and noise, but when sampled down to usable resolution they look great. Here's a sample of 1 Minimum Sample, 0 Adaptive Sampling and no noise reduction.
    Click image for larger version. 

Name:	EndTable(Med).jpg 
Views:	59 
Size:	901.5 KB 
ID:	146112
    Last edited by Tim Parsons; 10-09-2019 at 09:21 PM.
    Tim Parsons
    Technical Designer
    Sauder Woodworking Co.

    http://www.sauder.com

  8. #8
    Registered User
    Join Date
    Mar 2018
    Location
    Perth, Australia
    Posts
    82
    Quote Originally Posted by Tim Parsons View Post
    Not that I am aware of. And the sample scenes in the content aren't really optimized for rendering from what I've seen. I've made my own start scenes for product shots, but they they wouldn't be good for much else. For interiors I start from scratch on each job and learn something each time. I also do quite a bit of experimentation as well, but if you really want to learn some good practices to get you off on the right foot invest in RebelHill's video series. It was money well spent for me. I'm experimenting with super hi-res rendering with no AA samples and the results are promising. Fair rendering times, some jaggy's and noise, but when sampled down to usable resolution they look great. Here's a sample of 1 Minimum Sample, 0 Adaptive Sampling and no noise reduction.
    Click image for larger version. 

Name:	EndTable(Med).jpg 
Views:	59 
Size:	901.5 KB 
ID:	146112
    Hi Tim,

    I have done some experimenting with this as well. I found that it sometimes works better but other times its better rendering at intended res with more samples.
    Can I ask how high you render your output and what settings (Reflection/Refraction/SSS as well as GI ray samples)? And to what resolution do you scale it down? Also how long did the original render take?

    I had to render maybe 4 or 6 times the resolution in order to get nice AA when resampling (Cubic).

    Your example looks awesome by the way, very clean indeed. Is the DOF done in post?

  9. #9
    Quote Originally Posted by RebelHill View Post
    There is such a way, and what you're doing is not it... Start with the camera, then lights, then the rest. You can, ofc, do it any way round, but doing it the correct way will result in the lowest achievable render time for a given level of quality. All other approaches will do nothing more than waste time rendering.

    And no, there is no typical render, and the settings arent 'very' situational, they're completely so... There is no basic 'standard' of settings that make a good starting point... it's nothing but a waste of time to think there is.

    The workflow IS the standard, you follow it exactly the same way, for every scene, every situation, and it works every time.
    After being shut down so firmly by such a respected member of the community, my gut instinct is to let this response effectively kill this thread, shut my amateur mouth, and just walk away with my head down...

    But I'm gonna risk stepping into the fire to try to get clarification. There's no question you know exactly what you're doing, your experience and work speaks for itself. But looking through the forums, there are clearly many people who don't follow the same workflow. Perhaps they are just doing things incorrectly, but then why are there such varying opinions?

    In particular, to quote the LW docs themselves:
    "but be aware that those Camera samples are multipliers for the other sample settings so be prepared to juggle otherwise you will end with longer than necessary render times. If there's no motion blur or depth of field in your scene, look at the other buffers before you increase Antialiasing samples. You can probably get a clean render without increasing them."
    This seems to reinforce what I had said about the camera samples being a multiplier to the others. So adjusting camera samples 1st means your other sample tweaks will affect render times exponentially more, right?

    Other professionals I talked to when I first started with Lightwave 9 swore by the process of pinpointing where the problem areas were and adjusting appropriate reflection or light samples before deciding to increase the camera samples as a multiplier. I'm not trying to stir the pot, and I have huge respect for your work and contributions to the LW community. I'm just trying to understand why the 'camera, lights, everything else' workflow you described is what you consider to be the only right way.

  10. #10
    Quote Originally Posted by peebeearr View Post
    Hi Tim,

    I have done some experimenting with this as well. I found that it sometimes works better but other times its better rendering at intended res with more samples.
    Can I ask how high you render your output and what settings (Reflection/Refraction/SSS as well as GI ray samples)? And to what resolution do you scale it down? Also how long did the original render take?

    I had to render maybe 4 or 6 times the resolution in order to get nice AA when resampling (Cubic).

    Your example looks awesome by the way, very clean indeed. Is the DOF done in post?
    Here are the settings for this render. The Transparency, Reflection and Refraction Recursion Limits are set higher than I normally would because of the glass vase. (Other render settings are pretty much where I start usually) Render time was 46min with an i7 8086. DOF in post with a quick fog rendering. I did no noise reduction in the render or post.

    Click image for larger version. 

Name:	settings.jpg 
Views:	29 
Size:	156.2 KB 
ID:	146116
    Tim Parsons
    Technical Designer
    Sauder Woodworking Co.

    http://www.sauder.com

  11. #11
    Quote Originally Posted by SilverFalcon View Post
    After being shut down so firmly by such a respected member of the community, my gut instinct is to let this response effectively kill this thread, shut my amateur mouth, and just walk away with my head down...

    But I'm gonna risk stepping into the fire to try to get clarification. There's no question you know exactly what you're doing, your experience and work speaks for itself. But looking through the forums, there are clearly many people who don't follow the same workflow. Perhaps they are just doing things incorrectly, but then why are there such varying opinions?

    In particular, to quote the LW docs themselves:
    "but be aware that those Camera samples are multipliers for the other sample settings so be prepared to juggle otherwise you will end with longer than necessary render times. If there's no motion blur or depth of field in your scene, look at the other buffers before you increase Antialiasing samples. You can probably get a clean render without increasing them."
    This seems to reinforce what I had said about the camera samples being a multiplier to the others. So adjusting camera samples 1st means your other sample tweaks will affect render times exponentially more, right?

    Other professionals I talked to when I first started with Lightwave 9 swore by the process of pinpointing where the problem areas were and adjusting appropriate reflection or light samples before deciding to increase the camera samples as a multiplier. I'm not trying to stir the pot, and I have huge respect for your work and contributions to the LW community. I'm just trying to understand why the 'camera, lights, everything else' workflow you described is what you consider to be the only right way.
    I could be wrong but I think what RebellHill's point is that you have to determine your Anti-Aliasing needs first. - What is the amount of the jaggy's are you willing to live with? - And that can be determined real early in the setup and should be because there is nothing you can do with any other setting within LW that can reduce the jaggy's. Don't take offense, we are all learning.
    Last edited by Tim Parsons; 10-10-2019 at 08:18 AM.
    Tim Parsons
    Technical Designer
    Sauder Woodworking Co.

    http://www.sauder.com

  12. #12
    Super Member Kryslin's Avatar
    Join Date
    Feb 2009
    Location
    Prescott, IA
    Posts
    1,519
    If you are using FiberFX, you will need extra anti-aliasing passes, sometimes 30%-50% more. There is no way around it, due to the ray-traced primitive method now used.

    Nothing you can add to a scene (namely, a denoising filter) will denoise FiberFX fibers (Note, you can render oversize, and re-scale the image, reducing the extra AA passes you need, but it still takes a long time to render).
    --------
    My Scripts for Lightwave
    Intel Core i7 960 @3.20 Ghz, 24 GB ram, EVGA 6GB GTX980Ti "Classified" driving 2 x HP LA2405.

  13. #13
    Registered User
    Join Date
    Mar 2018
    Location
    Perth, Australia
    Posts
    82
    Quote Originally Posted by Tim Parsons View Post
    Here are the settings for this render. The Transparency, Reflection and Refraction Recursion Limits are set higher than I normally would because of the glass vase. (Other render settings are pretty much where I start usually) Render time was 46min with an i7 8086. DOF in post with a quick fog rendering. I did no noise reduction in the render or post.

    Click image for larger version. 

Name:	settings.jpg 
Views:	29 
Size:	156.2 KB 
ID:	146116
    Thanks Tim,

    That's indeed a pretty big image to render but the render time is pretty decent!
    I'm fooling around with the Intel denoiser quite a bit and get some nice results (for stills). But they are not as crisp as a native render unless you start cranking up resolution and samples. And then your back to square one :-) I'll try this and see how I go, thanks.

    But I'm gonna risk stepping into the fire to try to get clarification.

    Quote Originally Posted by SilverFalcon View Post
    After being shut down so firmly by such a respected member of the community, my gut instinct is to let this response effectively kill this thread, shut my amateur mouth, and just walk away with my head down...

    But I'm gonna risk stepping into the fire to try to get clarification. There's no question you know exactly what you're doing, your experience and work speaks for itself. But looking through the forums, there are clearly many people who don't follow the same workflow. Perhaps they are just doing things incorrectly, but then why are there such varying opinions?

    In particular, to quote the LW docs themselves:
    "but be aware that those Camera samples are multipliers for the other sample settings so be prepared to juggle otherwise you will end with longer than necessary render times. If there's no motion blur or depth of field in your scene, look at the other buffers before you increase Antialiasing samples. You can probably get a clean render without increasing them."
    This seems to reinforce what I had said about the camera samples being a multiplier to the others. So adjusting camera samples 1st means your other sample tweaks will affect render times exponentially more, right?

    Other professionals I talked to when I first started with Lightwave 9 swore by the process of pinpointing where the problem areas were and adjusting appropriate reflection or light samples before deciding to increase the camera samples as a multiplier. I'm not trying to stir the pot, and I have huge respect for your work and contributions to the LW community. I'm just trying to understand why the 'camera, lights, everything else' workflow you described is what you consider to be the only right way.
    SilverFlacon, there is no right or wrong really...(maybe a bit of waisted minutes... ) It kind of depends on what you are comfortable with. Rebels approach is very methodical. He really understands the background of settings. Why they are there, what they do and how they work. This is how he came up with a workflow that is based on proper knowledge. That helps in getting predictable results and will save you time in the end.

    Myself however, I like to fiddle around with settings and see what the impact is. This lead to many experiments and, as I said before, I do a lot now with OIDN. Tim's approach I have tried in the passed but stopped it quite quickly as I did not get good results. I have to say this was in the days before the new renderer... Now things are totally different and Tim has sparked my curiosity to start doing some tests with this again.

    In the end it's what you are most comfortable with. If everyone would go for the most effective way possible, the world would be pretty boring. Experiment by all means and let there be creativity

  14. #14
    RETROGRADER prometheus's Avatar
    Join Date
    Aug 2003
    Location
    sweden stockholm
    Posts
    15,100
    Quote Originally Posted by gar26lw View Post
    there are render presets, right?

    should get a set going for starting points. Maybe a set of example scenes would be better?
    Not sure if OD tools has something for that?
    Older lightwave had free plugin from Johan Steen, but I do not know if it works on 2019 since I havenīt tried it.
    https://www.lightwave3d.com/assets/p...ender-presets/

    Would be nice if we otherwise could specify favourite scenes in a load scene list or something..so you can bookmark your desired scenes to be reloaded as needed, if you want to start with a box, a volumetric primitive already setup etc.
    Sure..it wouldnīt be much of a trouble maybe to have scenes saved as preview, final render etc..in a dedicated start content folder, but anything that could be called upon within lightwave would make it much smoother and faster rather than browsing through folders with windows explorer.

    So Yes..set of scenes may be better..but as mentioned above, we should really be able to just pick bookmark favorites in a load preset scene shelf.

  15. #15
    Goes bump in the night RebelHill's Avatar
    Join Date
    Nov 2003
    Location
    jersey
    Posts
    5,770
    Quote Originally Posted by SilverFalcon View Post
    there are clearly many people who don't follow the same workflow.
    And those people are in all likelihood getting sub optimal rendertimes.

    Quote Originally Posted by SilverFalcon View Post
    then why are there such varying opinions?
    Why are there such varying opinions on whether the earth is round or flat? Its a mystery.

    Quote Originally Posted by SilverFalcon View Post
    to quote the LW docs themselves... This seems to reinforce what I had said about the camera samples being a multiplier to the others.
    Then the docs are wrong... or at the very least misleading.

    Camera samples are indeed a multiplier for all others... but they also are eneded to clean up camera level sampling effects. Therefore, you MUST turn up camera samples to raise these effects to the desired level of quality, which thus determines the MINIMUM number of camera samples which MUST be used... and since they are multipliying other effects, you clearly want to minimise them... so you start there because that is your unavoidable baseline.

    Quote Originally Posted by SilverFalcon View Post
    So adjusting camera samples 1st means your other sample tweaks will affect render times exponentially more, right?
    Nope, otherway round.

    Quote Originally Posted by SilverFalcon View Post
    I'm just trying to understand why the 'camera, lights, everything else' workflow you described is what you consider to be the only right way.
    Because there's no magic at work... It's not really about the "right" way, because there isn't one. Does your render have the level of quality/low noise desired... great it's "right"... but that's not the same thing as optimal in terms of the time taken to render.

    Many folk have this idea that there's some black magic, or secret sauce to the sample settings for this kind of scene, or that kind, that if you keep fiddling, fiddling, trying it this way, trying it that, that you can somehow hit upon the "ultimate tweak"... and it's just not true, it's a myth.

    There is, ofc, for any given scene, rendered to a given level of quality/noise reduction, one particular combination of the sample values which will yield a faster render than any other combination... and this magic number can be determined very easily, and very precisely by following one single workflow, every time, for every scene... it never fails. Ever.
    LSR Surface and Rendering Tuts.
    RHiggit Rigging and Animation Tools
    RHA Animation Tutorials
    RHR Rigging Tutorials
    RHN Nodal Tutorials
    YT Vids Tuts for all

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •