Messin' with Mhaze's Rock Study

Started by RichTwo, August 21, 2019, 11:49:18 am

Previous topic - Next topic

Dune

Users should make their decision based on knowledge about what each render method does and costs. Maybe most aren't completely incompetent, but some may just hit the button and start waiting...............................................................

WAS

Quote from: Dune on August 23, 2019, 03:00:24 amUsers should make their decision based on knowledge about what each render method does and costs. Maybe most aren't completely incompetent, but some may just hit the button and start waiting...............................................................

At which point logical reasoning may tell them it may be a long wait (...) at that bucket-rate, and they may need to reconsider their quality decisions, etc. I can see when buckets are slow and something needs to be changed if I'm around. Only time things are a issue is when I go to bed without monitoring and wake to 14 buckets finished of 48 after 8-12 hours.

Arbitrary learning can lead to some cool realizations but shouldn't be like a basis for advise. Yes PT is slow but it has it achieves it's goal.

And issues like this scene I feel it's a quirk of the current state of the PT rather than it as a whole or what to expect from a finished product. Such as rogue inter-reflective scenes from one angle to another creating a 800x render difference. Definitely not normal.

And while I may not have 12 cores to render in whole faster, my cores are clocked higher than a lot of users her and Inherently on their per-core performance, out-perform most workstation cores. 3.8ghz - 4.5ghz is pretty fast for a bucket. 
Check out my Terragen Discord: https://discord.gg/Vy5FRTE

RichTwo

Dune & WAS - I agree with you both. PT may not be *required* but produces *desirable* results. And no it's not for everyone.  Especially someone like me looks for an easy way out, and doesn't want to wait long for it. Not for very long anyway. I have another version in the works, no PT only turning down the Enviro Lighting to just under 1.0.  Test render was only about an hour which to me is reasonable.
You don't need a weatherman to know which way the wind blows...

WAS

Quote from: Rich2 on August 23, 2019, 12:29:00 pmDune & WAS - I agree with you both. PT may not be *required* but produces *desirable* results. And no it's not for everyone.  Especially someone like me looks for an easy way out, and doesn't want to wait long for it. Not for very long anyway. I have another version in the works, no PT only turning down the Enviro Lighting to just under 1.0.  Test render was only about an hour which to me is reasonable.


Of course not but the entire argument laid out is invalid and based in a falsifiability. No PT doesn't just effect foliage shadows, and no it's not only really good for lots of foliage. It is good for everything. Whether you use it or not is not even the concern here. It's how you spread information about it as advise, as superior. We have enough gripe about available features that doesnt effect the community or software in a positive manner. Need to stop saying stuff like that and let people do them, and make their own decisions.

The convo here about PT is that it shouldn't have taken that long for your render, and really should be investigated. There is nothing in the scene that should cause such a slowdown unless it's how PT is handling the background node or something.
Check out my Terragen Discord: https://discord.gg/Vy5FRTE

Oshyan

August 23, 2019, 09:25:47 pm #19 Last Edit: August 23, 2019, 10:22:14 pm by Oshyan
Quote from: WAS on August 23, 2019, 03:21:33 amAnd issues like this scene I feel it's a quirk of the current state of the PT rather than it as a whole or what to expect from a finished product. Such as rogue inter-reflective scenes from one angle to another creating a 800x render difference. Definitely not normal.
The issue you originally reported seemed to be about a 27x difference (although in that post you said "10000x"). But you have since identified an antivirus issue that was surely affecting the result, not to mention it was also not reproducible on our systems here, nor have we had any other reports of such issues. The render time would understandably be longer at the different angle due to inter-reflection, but it was nowhere near "800x" (nor 10000x). Render times change based on many factors, including camera angle, and regardless of PT. Please stop spreading misinformation.

Furthermore, while PT is not exclusively beneficial for scenes with objects, plants, etc., it is fair to say that the results are often (but not exclusively) more noticeable in such scenarios. This is simply because the standard renderer handles large-scale and simple light bouncing and occlusion calculations fairly well. Simple mountain, desert, rock, etc. scenes are easier for cache-based methods to do a decent job on. More complex lighting scenarios that result from multiple layers of overlapping geometry, occlusion, shadowing, etc. will tend to show the results of path tracing more. So the advice that Ulco is giving is perhaps too strongly stated, but it is not completely wrong. Simply put, the render time difference with PT on vs. off is more likely to be worth it (i.e. to make a big difference) in more complex situations.

In the example in this scene here I'm not even sure I believe that the PT on/off comparison is quite valid because the difference is so dramatic. I wouldn't jump to conclude that the difference is due solely to PT. For example the Enviro Light has no effect with PT enabled, so if in the original the enviro light was increased (as it sounds like it might have been), then the PT version would be inherently darker, not due to being more accurate, but simply to essentially disregarding the enviro light increase. Without knowing the rest of the scene settings it's just hard to know if it's a good comparison.

That is not to say such comparisons are not worthwhile and interesting, nor that trying PT on a scene like this is bad idea. I'm only cautioning against drawing major conclusions about PT vs. non based on comparisons like this. It will take a while before everyone fully understands the specific consequences of Terragen's path tracing mode and settings, and can thus make accurate comparisons. And we'll do our best to document and make people aware of things that might make comparison harder, like the enviro light difference.

As for the much longer render time, without knowing the full settings of the PT render I don't know if it "should" take that long or not. PT requires optimization of settings, which I've outlined elsewhere (and, yes, will be added to the docs, with even more detail). We are of course happy to work with Rich or anyone else who wants help optimizing render settings and time, that just requires sharing scene files. We'd welcome Rich getting in touch if he's interested in doing that.

- Oshyan

WAS

August 23, 2019, 11:24:36 pm #20 Last Edit: August 23, 2019, 11:32:29 pm by WAS
Quote from: Oshyan on August 23, 2019, 09:25:47 pm
Quote from: WAS on August 23, 2019, 03:21:33 amAnd issues like this scene I feel it's a quirk of the current state of the PT rather than it as a whole or what to expect from a finished product. Such as rogue inter-reflective scenes from one angle to another creating a 800x render difference. Definitely not normal.
The issue you originally reported seemed to be about a 27x difference (although in that post you said "10000x"). But you have since identified an antivirus issue that was surely affecting the result, not to mention it was also not reproducible on our systems here, nor have we had any other reports of such issues. The render time would understandably be longer at the different angle due to inter-reflection, but it was nowhere near "800x" (nor 10000x). Render times change based on many factors, including camera angle, and regardless of PT. Please stop spreading misinformation.

Furthermore, while PT is not exclusively beneficial for scenes with objects, plants, etc., it is fair to say that the results are often (but not exclusively) more noticeable in such scenarios. This is simply because the standard renderer handles large-scale and simple light bouncing and occlusion calculations fairly well. Simple mountain, desert, rock, etc. scenes are easier for cache-based methods to do a decent job on. More complex lighting scenarios that result from multiple layers of overlapping geometry, occlusion, shadowing, etc. will tend to show the results of path tracing more. So the advice that Ulco is giving is perhaps too strongly stated, but it is not completely wrong. Simply put, the render time difference with PT on vs. off is more likely to be worth it (i.e. to make a big difference) in more complex situations.

In the example in this scene here I'm not even sure I believe that the PT on/off comparison is quite valid because the difference is so dramatic. I wouldn't jump to conclude that the difference is due solely to PT. For example the Enviro Light has no effect with PT enabled, so if in the original the enviro light was increased (as it sounds like it might have been), then the PT version would be inherently darker, not due to being more accurate, but simply to essentially disregarding the enviro light increase. Without knowing the rest of the scene settings it's just hard to know if it's a good comparison.

That is not to say such comparisons are not worthwhile and interesting, nor that trying PT on a scene like this is bad idea. I'm only cautioning against drawing major conclusions about PT vs. non based on comparisons like this. It will take a while before everyone fully understands the specific consequences of Terragen's path tracing mode and settings, and can thus make accurate comparisons. And we'll do our best to document and make people aware of things that might make comparison harder, like the enviro light difference.

As for the much longer render time, without knowing the full settings of the PT render I don't know if it "should" take that long or not. PT requires optimization of settings, which I've outlined elsewhere (and, yes, will be added to the docs, with even more detail). We are of course happy to work with Rich or anyone else who wants help optimizing render settings and time, that just requires sharing scene files. We'd welcome Rich getting in touch if he's interested in doing that.

- Oshyan

There have been reports all over the forums about slow PT render times and opting for SR, leading to conversations like try DA or just use SR.

Definitely not even going to touch your further falsifiabilities about Path Tracing and why it's a moving standard in the industry. Has nothing to do with vegetation or shadows. That being said here is negligent as all hell Oshyan. If that's Matt's focus on PT it should be elaborated on as it's not PTs goal in the industry. It's for physically correct surfaces. For example PBR Materials conversation. They are for PT renderers. This includes shadows of course but is certainly not a key point. Not even used in documentation about PT that's ive come across. It's about surfaces and lighting in general. Why you feel the need to defend such a falsifiability is beyond me. Lmao

If you can't see the differences in PT scenes and non PT scenes perhaps you're just not that sensitive to art critique or subtle changes in detail. But they're plain as day to me. I feel we get into this arguments a lot where reference photos need to be dropped and than you argue invalid observations like lighting and specular of surfaces like cement vs "glass-like" obsidian with full reflection.
Check out my Terragen Discord: https://discord.gg/Vy5FRTE

Matt

Quote from: WAS on August 23, 2019, 11:24:36 pmDefinitely not even going to touch your further falsifiabilities about Path Tracing and why it's a moving standard in the industry. Has nothing to do with vegetation or shadows. That being said here is negligent as all hell Oshyan.

Oshyan's statements seem quite reasonable to me. He's pointing out why sometimes you don't need PT in a Terragen scene, and it's useful to know because non-PT is faster. Sure, PT looks better and he never claimed otherwise. How can you call any of what he said a "falsifiability" or "negligent as all hell" and expect us to take you seriously?
Just because milk is white doesn't mean that clouds are made of milk.

WAS

August 23, 2019, 11:58:18 pm #22 Last Edit: August 24, 2019, 12:01:42 am by WAS
Quote from: Matt on August 23, 2019, 11:50:52 pm
Quote from: WAS on August 23, 2019, 11:24:36 pmDefinitely not even going to touch your further falsifiabilities about Path Tracing and why it's a moving standard in the industry. Has nothing to do with vegetation or shadows. That being said here is negligent as all hell Oshyan.

Oshyan's statements seem quite reasonable to me. He's pointing out why sometimes you don't need PT in a Terragen scene, and it's useful to know because non-PT is faster. Sure, PT looks better and he never claimed otherwise. How can you call any of what he said a "falsifiability" or "negligent as all hell" and expect us to take you seriously?


There is no contest about which is faster. My input is about quality difference. If that is a concern, and you want more physically correct renders, you'd of course go for what? A PT render.

You can, again, see the difference in any TG scene just about. You can see the difference like night and day through difference (layer) comparisons between highights, subtle shadows, and shadows, that's all the scene. Not to say you can't fake a somewhat PT look in SR by using reflectivity on all surfaces to a subtle degree which I've done for years, but that's just as slow almost.

Just like the differences are plain as day in Rich's.

I was never arguing that you shouldn't use SR but the reason not to laid  here by Dune and enforced by Oshyan, a BM of PS, is plain wrong. It's not "only really good for vegetation". Period.
Check out my Terragen Discord: https://discord.gg/Vy5FRTE

Matt

@WAS, he clearly never said that. If you want to talk about "falsifiabilities" (your word, not mine), please look at your own statements first.

Look what we've done to this thread. Can we get back on topic now?
Just because milk is white doesn't mean that clouds are made of milk.

WAS

August 24, 2019, 12:29:58 am #24 Last Edit: August 24, 2019, 12:35:46 am by WAS
Quote from: Matt on August 24, 2019, 12:13:01 am@WAS, he clearly never said that. If you want to talk about "falsifiabilities" (your word, not mine), please look at your own statements first.

Look what we've done to this thread. Can we get back on topic now?

Seriously?

Quote from: DunePT is really only very good for landscapes with lots of foliage
(Than argued)

Quote from: OshyanFurthermore, while PT is not exclusively beneficial for scenes with objects, plants, etc., it is fair to say that the results are often (but not exclusively) more noticeable in such scenarios...

Where's is that not comprehended as what it is? Especially when Oshyan goes on to specifically say it's not for really good for landscapes. Lol Like jesus.  And I gotta add again, when you go and read documentation and articles the focal point here at PS about vegetation isn't even a focal point or what's demonstrated anywhere else for Path Tracing. It's about how surfaces, shadows, lighting, etc. This of course, again, includes vegetation but why it's such a focus here here beyond me. No other engine or article/docu on the method does.

And yes we are blowing up this port image showcases thread.
Check out my Terragen Discord: https://discord.gg/Vy5FRTE

Matt

Quote from: WAS on August 24, 2019, 12:29:58 am
Quote from: DunePT is really only very good for landscapes with lots of foliage
(Than argued)

Dune said this, not Oshyan.

Quote from: WAS on August 24, 2019, 12:29:58 am
Quote from: OshyanFurthermore, while PT is not exclusively beneficial for scenes with objects, plants, etc., it is fair to say that the results are often (but not exclusively) more noticeable in such scenarios...

Where's is that not comprehended as what it is?

The entire sentence. It doesn't say PT is only good for vegetation. If you think it does, then I'm afraid you misunderstood the sentence.
Just because milk is white doesn't mean that clouds are made of milk.

WAS

August 24, 2019, 01:18:03 am #26 Last Edit: August 24, 2019, 01:21:12 am by WAS
Quote from: Matt on August 24, 2019, 12:57:14 am
Quote from: WAS on August 24, 2019, 12:29:58 am
Quote from: DunePT is really only very good for landscapes with lots of foliage
(Than argued)

Dune said this, not Oshyan.

Quote from: WAS on August 24, 2019, 12:29:58 am
Quote from: OshyanFurthermore, while PT is not exclusively beneficial for scenes with objects, plants, etc., it is fair to say that the results are often (but not exclusively) more noticeable in such scenarios...

Where's is that not comprehended as what it is?

The entire sentence. It doesn't say PT is only good for vegetation. If you think it does, then I'm afraid you misunderstood the sentence.


I think you're clearly misunderstanding what clearly being conveyed. He literally verbatim continues to go on to single out landscapes (the target discussion over shadow sampling of vegetation he was clearly discussing).

Wooow.

It's clear he's reinforcing the narrative about PT good for vegetation and shadows.

Seems you're either not understanding what's clearly being conveyed or trying to play a weird deflection on technicalities because he didn't say it specifically by name "vegetation".  He was explaining why it's good for shadows, and why it's not really "needed" for landscapes, which I entirely disagree with and can see the benefits plain as day. Doesn't mean I'll always use it, cause, again, speed. But it is ofc better than SR.
Check out my Terragen Discord: https://discord.gg/Vy5FRTE

Dune

I should have expected such WAS-on-the-loose replies, sorry guys. I'm not saying much more than what I meant was just that PT is very good for landscapes with lots of vegetation, and good for landscapes without much vegetation, versus non-PT (choice of course taking into account the time a render would take in either setting). That's all. I won't say stuff like that anymore :-X

Matt

Quote from: WAS on August 24, 2019, 01:18:03 amSeems you're either not understanding what's clearly being conveyed or trying to play a weird deflection on technicalities because he didn't say it specifically by name "vegetation".  He was explaining why it's good for shadows, and why it's not really "needed" for landscapes, which I entirely disagree with and can see the benefits plain as day. Doesn't mean I'll always use it, cause, again, speed. But it is ofc better than SR.

There is no one single answer to how important PT is to landscapes. I have my views on the importance of PT, but they have to be tempered with the practical implications of rendering that way. Yes, I am getting into technicalities here because you said he was "negligent as all hell" and that what he said was full of "falsifiabilities". Ironically you then claimed he said things he didn't, which is just asking to be falsified. This is the first time I've done this on the forum because I have a lot of respect for the people here and we don't usually talk to each other like this. Do you want to continue or shall we stop it here?
Just because milk is white doesn't mean that clouds are made of milk.

Matt

Quote from: Dune on August 24, 2019, 01:56:09 amI should have expected such WAS-on-the-loose replies, sorry guys. I'm not saying much more than what I meant was just that PT is very good for landscapes with lots of vegetation, and good for landscapes without much vegetation, versus non-PT (choice of course taking into account the time a render would take in either setting). That's all. I won't say stuff like that anymore :-X

Please don't stop saying anything @Dune! We learn a lot from you. Sometimes people will disagree with you, but that's OK! :)
Just because milk is white doesn't mean that clouds are made of milk.