Somebody buy the iray speed tutorial and then explain what it is.

2

Comments

  • 3Diva3Diva Posts: 11,970
    edited November 2017
    avxp said:

    Well done Jaxprog. We need more threads like this- with helpful insights pros/cons around products with vague descriptions and such.

    +1 I'd love to see more product review threads. I know that over in the art studio section @Novica has a really nice product review thread. I'd also like to see some reviews from others as well. :)

    Post edited by 3Diva on
  • Well, I'd possibly like (and it's a can of worms) but just how there's a commercial release official thread, there could also be a Product Review thread, like a section.

    And you do the Amazon thing, whee it has to be a certified purchase...And writinga  review could have a submission form with criteria so it's not just knee-jerk responses.

    Or do official review postings. And then USERS get comment down after.

    I would thing, just the opening paragraph of the review would give sooo many details, it's be worth reading.

    Let the reviewer explain what it is first. And mods (if they chose to) could sort the Top/Pinned review so that the most professional comprehensive review remains the first one you read.

  • NovicaNovica Posts: 23,924
    avxp said:

    Well done Jaxprog. We need more threads like this- with helpful insights pros/cons around products with vague descriptions and such.

    +1 I'd love to see more product review threads. I know that over in the art studio section @Novica has a really nice product review thread. I'd also like to see some reviews from others as well. :)

    Why thank you!  Although it started with me doing the reviews, it expanded about three years ago to include forum members reviewing products. That way it's all centrally located.  So hasn't been "me" reviewing for quite awhile (thank goodness, couldn't keep up!)

    FYI I also just linked a Content Package Assist tutorial from that thread (in addition to products, we discuss sales (five years longer than the Commons sale thread, I might add) which includes products on sale that we like, and tips/tutorials. 

     I do ask that people doing reviews be courteous, factual, and have good renders to show the products in a favorable manner. (No dark renders lacking detail or overlit renders with high glare/burnout, no WIPS that need "fixing." In other words, polished renders.) The renders are also family friendly (can be sexy, but not exaggerated.) The best idea is to surf the thread to get an idea of the "culture" of it. Love to have more contributing reviewers! 

  • BlueIreneBlueIrene Posts: 1,318

    Having seen the positives mentioned by others, I bought the product. I'm pretty certain I've implemented all of the more useful (to me) tips at one time or another in the past and still use some of them now, although I don't always think to use the others and had even forgotten that I knew some of them. I'll be using the first one regarding render settings all the time in future - now that it's been explained to me and I know what I'm aiming for, I've been experimenting with it tonight and it does make a substantial difference, at least on my machine. I never once felt insulted, even when I encountered a video that demonstrated what I already knew. After a few years at this there would be something very wrong if all of it was new to me. Only one video had me thinking 'You're kidding. Could you possibly get more obvious?' at the beginning, but it went off in a different direction to what I'd expected and I finished the video thinking 'I'll definitely be giving that one a go when it fits in with what I'm doing'.

    If there was anything 'never before released' or 'shocking' in there then I was oblivious to it, but there was enough in there for me to make it a worthwhile purchase, particularly at the current price.

  • davesodaveso Posts: 7,784

    I'm using an onboard, non nvidia GPU, so iray is cpu rendered. will these tips help at all or would I be wasting my money and maybe experiment with render settings instead. I need to buy a dedicated graphics card someday, but they are way too expensive, and the ones that I could afford really dont up the game all that mcuh. 

  • 3Diva3Diva Posts: 11,970

    Having seen the positives mentioned by others, I bought the product. I'm pretty certain I've implemented all of the more useful (to me) tips at one time or another in the past and still use some of them now, although I don't always think to use the others and had even forgotten that I knew some of them. I'll be using the first one regarding render settings all the time in future - now that it's been explained to me and I know what I'm aiming for, I've been experimenting with it tonight and it does make a substantial difference, at least on my machine. I never once felt insulted, even when I encountered a video that demonstrated what I already knew. After a few years at this there would be something very wrong if all of it was new to me. Only one video had me thinking 'You're kidding. Could you possibly get more obvious?' at the beginning, but it went off in a different direction to what I'd expected and I finished the video thinking 'I'll definitely be giving that one a go when it fits in with what I'm doing'.

    If there was anything 'never before released' or 'shocking' in there then I was oblivious to it, but there was enough in there for me to make it a worthwhile purchase, particularly at the current price.

    Thank you for the review, Astracadia! The more more I hear about it, the more I'm tempted to get it. I've only been doing 3D art for less than 2 years, so I figure I could probably pick up some helpful tips on how to decrease my render times.

  • GoggerGogger Posts: 2,487

    One thing I have learned with Dreamlight tutorials is that no matter what I thought I was going to get out of it, I invariably DO get *something* that ends up making a BIG difference in how I do art and these things stick with me and improve my work from that point on. Sure, it would be nice if ALL the things promised happened, but I feel pretty much guaranteed that I will get something that will improve my work.

    That said, I was going to pass on this but after reading through this thread, decided to go ahead and pick it up tonight.  I'll try to post my thoughts after I get a chance to view it.

  • HorusRaHorusRa Posts: 1,664
    edited March 2019

    .

    Post edited by HorusRa on
  • Oh, they may be widely known, but from comments made regularly on this forum. they are not used, as shown by all the complaints of render times. I've been doing a variation on the render settings for some time now, and I normally have renders on my old 780 Ti in several minutes to 20 minutes. And I often suggest compositing tips yet people resist.

  • Hmm, a thread about render speed and quality yet NO images have been provided by anyone. Stop talking and start showing. Pictures pictures pictures!!!! What we need to do is to post a render of a scene as the artist had originally set it up followed by a second render using the concepts and tools provided by the videos. Let us know the render times, let us observe for ourselves that the second one takes less than half the time of the first one. That then allows us to decide if the information in these tips provides any feasible boost in speed and or quality, or if these tips are in some way detrimental. There is no need to describe EXACTLY which tips were used, only for people to see that quality can be maintained to a certain degree while increasing speed for those who need it by implementing these settings and tricks. Just show us the images.

    The advice I've read so far seems useful indeed. However i can already disagree with some of it. I personally have seen the advice to rendering larger and sampling down offered in many forums regarding many software packages as a way to speed rendering, but I cannot for the life of me figure out why. If that "trick" worked as well as people think it does, then why is this behavior not a default behavior of any rendering engine on the market? Not one programming team has convinced their bosses that adding this functionality to their renderer will speed rendering with absolutely no loss in quality. If I am wrong I'd gladly appreciate being corrected.

    I'm also not in love with certain compositing ideas, however I do know that this is very much the way things are done in professional studios so I remain open to such things. True, it spoils the lack of biasing in the rendering but then again who really cares if the image ends up close enough to the correct values?

  • BlueIreneBlueIrene Posts: 1,318
    The advice I've read so far seems useful indeed. However i can already disagree with some of it. I personally have seen the advice to rendering larger and sampling down offered in many forums regarding many software packages as a way to speed rendering, but I cannot for the life of me figure out why. If that "trick" worked as well as people think it does, then why is this behavior not a default behavior of any rendering engine on the market? Not one programming team has convinced their bosses that adding this functionality to their renderer will speed rendering with absolutely no loss in quality. If I am wrong I'd gladly appreciate being corrected.

    I haven't got time to set up and render a 'tips' version of an image today, let alone the tips version and the equivalent long version as originally envisaged (I get why you're asking though and hope someone with time to spare today or a faster machine supplies it for you). However, I just thought I'd say that rendering larger and sampling down isn't among the suggestions the product makes. I've never been keen on that technique either, but was a bit surprised that the spot render tool didn't get a mention too (unless I've already forgotten it!). That's something I use all the time to focus on and quickly get rid of big noise patches when the rest of the image is already cooked.

  • Pictures pictures pictures!!!! What we need to do is to post a render of a scene as the artist had originally set it up followed by a second render using the concepts and tools provided by the videos.

    I'm not sure how much use that would actually be. Remember that Iray rendering is identical no matter what device it's rendered on — assuming that a slow CPU render hasn't been stopped prematurely by hitting the time limit. Everything else being equal, the two renders ought to be exactly the same.

  • Kevin SandersonKevin Sanderson Posts: 1,643
    edited November 2017

    You find the setting that will work for the render in speeding it up. They could/will be different. It's not one setting fits all. You have to experiment. You find what's good enough or looks pretty good.

     

    Test1.jpg
    1920 x 1080 - 478K
    Test2.jpg
    1920 x 1080 - 491K
    Post edited by Kevin Sanderson on
  • bluejauntebluejaunte Posts: 1,990

    You find the setting that will work for the render in speeding it up. They could/will be different. It's not one setting fits all. You have to experiment. You find what's good enough or looks pretty good.

     

    2nd render quite a bit more grainy though.

  • I like it when people review a product. I am a big fan of amazon's customer reviews it gives me guidance if to purchase or not. I will give this tutorial a second look because of the review. Thank you Jaxprog.

  • Kevin SandersonKevin Sanderson Posts: 1,643
    edited November 2017

    You find the setting that will work for the render in speeding it up. They could/will be different. It's not one setting fits all. You have to experiment. You find what's good enough or looks pretty good.

     

    2nd render quite a bit more grainy though.

    If you hadn't seen the first, you wouldn't notice. I think it's only a little grainier.  Of course it rendered much faster, 8 minutes compared to 21 minutes. An intermediate time rendering, and you probably couldn't tell otherwise. A good test to try is getting a frame from an actual movie and see how grainy that is. Then reality sets in. Some times I think people here are perfectionists to the extreme. And sadly it's holding them back. I know of one in particular who's always bemoaning his lack of equipment  but absolutely refuses to composite. And having seen what he has done in the past, he would get very close results if he did, and be able to use his older gear. When I was a radio production director the urge was to get everything sounding as pristine and perfect as possible. But in the time available and workload present, you couldn't do it, so you end up trying the best you can. Most people would never be able to tell. To this day as I do traffic reports, I'll pick up my own mistakes, but listeners and co-workers never hear them. I think most people would never notice the fine points of many renders as well.



    The time savings are true, and it sure beats the several hours some are spending.

    Post edited by Kevin Sanderson on
  • bluejauntebluejaunte Posts: 1,990
    edited November 2017

    Sure, was merely pointing out that it's more grainy so less render time than the first image may not be down just to superior settings but to... well, less render quality :)

    Post edited by bluejaunte on
  • BlueIreneBlueIrene Posts: 1,318
    edited November 2017

    These are only 500 x 650 as I haven't got time for anything bigger, but the first took 36 minutes and 5 seconds on Old Steam-driven, and the second version - using the render settings 'trick' described in the product - took 12 minutes and 4 seconds. A third of the time. There's a little extra noise in the open doorway and in the shadow of the tree, and I could experiment with the settings to find a sweet spot that would get rid of that while still massively cutting the render time, but I'd probably just give these bits a quick blast with the spot render tool instead using a few of those 24 minutes I've saved. Life's too short.

    Original.png
    500 x 650 - 512K
    Tweaked.png
    500 x 650 - 520K
    Post edited by BlueIrene on
  • CypherFOXCypherFOX Posts: 3,401
    edited November 2017

    Greetings,

    Hmm, a thread about render speed and quality yet NO images have been provided by anyone. Stop talking and start showing. Pictures pictures pictures!!!! What we need to do is to post a render of a scene as the artist had originally set it up followed by a second render using the concepts and tools provided by the videos. Let us know the render times, let us observe for ourselves that the second one takes less than half the time of the first one. That then allows us to decide if the information in these tips provides any feasible boost in speed and or quality, or if these tips are in some way detrimental. There is no need to describe EXACTLY which tips were used, only for people to see that quality can be maintained to a certain degree while increasing speed for those who need it by implementing these settings and tricks. Just show us the images.

    The whole point of a tips video like that is that the result is near or exactly identical to what you would get with a slower render time.  I don't think before/after are necessarily good ideas in this case.  I fear folks nitpicking and losing sight of the broader point.  But that's how things go, so... ¯\_(ツ)_/¯

    The advice I've read so far seems useful indeed. However i can already disagree with some of it. I personally have seen the advice to rendering larger and sampling down offered in many forums regarding many software packages as a way to speed rendering, but I cannot for the life of me figure out why. If that "trick" worked as well as people think it does, then why is this behavior not a default behavior of any rendering engine on the market? Not one programming team has convinced their bosses that adding this functionality to their renderer will speed rendering with absolutely no loss in quality. If I am wrong I'd gladly appreciate being corrected.

    Sure, I'll take a shot at explaining this, because this is a tip I use fairly frequently.  You can't use this tip on render engines that go pixel by pixel (raytracing engines like 3Delight).  Engines like LuxRender and Iray will benefit from this because of the way they scatter light onto the resultant canvas.  The core idea here is that if you have plenty of time to render, then you don't use this tip, because it won't make a lot of difference.

    The idea is that rendering at 2560x1440px and downsampling to 1280x720px (using a good image editing program) gives you 2x anti-aliasing.  That is to say 2 pixels in each direction is used to enhance the final pixel.  So four pixels combine to each point, which means you're essentially quadrupling your sampling at each resultant point.  This is important because often the fireflies or noise in an image are just one random pixel that gets a weird value (one particular light path scatters oddly and brightens a pixel that it shouldn't).  The oversampling reduces the strength of those random scatters, making the ultimate imge more noise-free.

    The only goal here is to reduce noise in a shorter time, and downsampling does help with that.  If you're rendering something overnight, or 'until it's done' with some sort of render convergence, then this tip won't be helpful.

    It's only useful in limited (but not rare) circumstances, when you have a limited time budget and noise concerns.  That's why it's not a standard in all appropriate engines, and why it's not something you always use.  A good example of a place to use this is in rendering for animation.

    It's not a cure-all, but there are no silver bullets.


    That all said, I admit the conversation here on this item has made me consider this tutorial more seriously.  I just wish he wouldn't make such absurdly grandiose claims in his product descriptions.  It leads to me, at least, taking him MUCH less seriously.

    --  Morgan

     

    Post edited by CypherFOX on
  • Kevin SandersonKevin Sanderson Posts: 1,643
    edited November 2017

    Here are 3 renders... one full default setting, an in between setting and the original lower setting. The in between setting still saved 6 minutes.

     

    Test1.jpg
    1920 x 1080 - 478K
    Test3_inbetweensettings.jpg
    1920 x 1080 - 369K
    Test2.jpg
    1920 x 1080 - 491K
    Post edited by Kevin Sanderson on
  • bluejauntebluejaunte Posts: 1,990

    Here are 3 renders... one full default setting, an in between setting and the original lower setting. The in between setting still saved 6 minutes.

     

    Ok, I have to ask. With each render you're going through less iterations. If you used the settings from the first render and stopped it at 8 mins, does it look worse than the last render?

  • Kevin SandersonKevin Sanderson Posts: 1,643
    edited November 2017

    No, it's not stopped. Those are 100% renders with different settings. Two settings are lower than the original.

     

     

    Post edited by Kevin Sanderson on
  • bluejauntebluejaunte Posts: 1,990

    Yeah, what I meant is if you did manually stop the first render at 8 mins. How would it look?

  • bluejaunte, here's one manually stopped with the default settings at 8 minutes 10 seconds to mirror Test2's time.

    Test4.manuallystopped.jpg
    1920 x 1080 - 646K
  • bluejauntebluejaunte Posts: 1,990

    bluejaunte, here's one manually stopped with the default settings at 8 minutes 10 seconds to mirror Test2's time.

    Ok, looks pretty much like render 3. I get it now, essentially you're not really optimizing anything, the changed settings simply let Iray run for less long before it's at 100%. I guess it's good for people who always let Iray render to 100%. This isn't needed of course, you can stop a render at any time if you're happy with what grain remains.

  • RhaleRhale Posts: 78

    You find the setting that will work for the render in speeding it up. They could/will be different. It's not one setting fits all. You have to experiment. You find what's good enough or looks pretty good.

     

    I like this Kevin - as an aside, what character/texture and hair is this?

  • bluejaunte, here's one manually stopped with the default settings at 8 minutes 10 seconds to mirror Test2's time.

    Ok, looks pretty much like render 3. I get it now, essentially you're not really optimizing anything, the changed settings simply let Iray run for less long before it's at 100%. I guess it's good for people who always let Iray render to 100%. This isn't needed of course, you can stop a render at any time if you're happy with what grain remains.

    I think that's the problem many have, they think they have to let it go all the way when several minutes will often do the job, and mine were done on an old 780 Ti. He has a Photoshop tip for reducing noise, but there are a couple other things you can do. I think some people are trying to get rid of noise with longer renders and that may not work some times. His light tip has something to do with that as well. His background idea has been around, but many around here don't seem to know it, and it can save big time. It factored into his earlier movie background kits that do render fast.

  • WonderlandWonderland Posts: 7,133

     Does this work on CPU only or do you need an NVIDIA card GPU? 

  • Kevin SandersonKevin Sanderson Posts: 1,643
    edited November 2017

    @Wonderland Val says over on facebook that the techniques will work for both Iray on Nvidia GPU and CPU, but I would guess a Nvidia GPU will be faster depending on your scene since it's normally faster than a CPU. But time savings are good any way you look at it. The actual product is a series of video tutorials that will run on any computer.

    @Rhale_2 Thanks! The character is a slightly modified Emma and Jordi Adora https://www.daz3d.com/ej-adora-for-genesis-3-female-s  (they have a version for G8F in a pack) and the hair is Anjou Hair by ApriYSH https://www.daz3d.com/anjou-hair

    Post edited by Kevin Sanderson on
  • He has a Photoshop tip for reducing noise,

    Does it work better than Filter>Noise>Reduce Noise for noise, and Filter>Noise>Dust and Scratches for fireflies?

Sign In or Register to comment.