Nearly Identical scenes, vastly different render times.

[slitherhence]

I have two renders/scenes.

Spoiler: unprocessed renders

You don't have permission to view the spoiler content. Log in or register now.

The scenes have the following 4 known differences:

1. The demoness' pose is different.
2. There is a "squirt" wearable added to the later scene. Confirmed not the problem... issue still occurred with it removed.
3. The demoness' tiara was morphed slightly in the latter image to not clip through her forehead. Confirmed not the problem... issue still occurred with morphs reset.
4. The former scene took ~1 hour 10 minutes and 3000+ iterations to render while the latter hit max render time at 2hours and only accomplished 230 iterations in that time.

There are no other known differences. I eliminated all elements of the set, render settings, hdri, lighting, camera, and everything else not part of the demoness as possible causes by saving her as a scene subset, loading the former scene, removing the demoness from it, and finally loading the saved scene subset. Thus, if there are any unknown differences they must be part of the demoness or items with the demoness as parent (clothing, hair, etc)... or must have been introduced by the act of removing one demoness model and loading another via scene subset. Unfortunately, that's a long list of possible differences to check for.

So... why does each iteration take so long when rendering the latter scene? Can anyone point me where to look for likely culprits? I can't wait 26 hours per render to achieve parity in the scenes here.... especially when there are eight other scenes in this set having the same issue.

I'm using the same RTX 3070 to render both scenes. Latest DAZ Studio. Iray.

EDIT: Because this seems to be a point of confusion for people... here is a render of the first scene artificially limited to 230 iterations (it took 10 minutes).

Spoiler: Intentionally hobbled render

You don't have permission to view the spoiler content. Log in or register now.

As you can see, when the scene that is rendering quickly is artificially limited to the same number of iterations the appearance of the skin is identical. The two figures in these scenes do not have different materials for the skin. I've manually confirmed that every single property of both skin materials are identical. The apparent difference between the skin and other materials in the scene is because the skin material is very complex. It is, after all, the main focus of the scene... as it is with any porn scene.

CPU rendering is already disabled. Has been pretty much since I started doing this. I have also confirmed in the logs that the iray renderer is using the nvidia GPU for rendering for the renders that are excessively slow. If iray is switching to CPU rendering it is doing so in spite of my settings and then lying about it in the log while also somehow tricking the GPU into showing a high load and high vram consumption during renders.

 

[anne O'nymous]

There are no other know differences.

Really ? When looking at the two renders, they don't seem seem to have the same skin materials.

Thus, if there are any unknown differences they must be part of the demoness or items with the demoness as parent (clothing, hair, etc)...

The clothing, hair, and nails materials are clearly the same, and already near to their final quality in the second render. The sole visible difference is the skin ; and I don't talk about it's color, it's more surely a consequence of this difference, than the cause.

 

[slitherhence]

Really ? When looking at the two renders, they don't seem seem to have the same skin materials.

The clothing, hair, and nails materials are clearly the same, and already near to their final quality in the second render. The sole visible difference is the skin ; and I don't talk about it's color, it's more surely a consequence of this difference, than the cause.

The skin materials are identical (as far as I can tell, i'll investigate further when I get back from lunch). What you are seeing if the result of it failing to produce enough iterations for the skin to be completed. Like I said, the first render took over 3000 iterations to reach 95% convergence after 1hour 10minutes. Meanwhile the second render never indicated even 1% convergence after 2 hours... and only completed 230 iterations. At 230 iterations the first render looked much the same... but 230 iterations for it took all of 10-15 minutes (including startup time).

If you zoom in you can very clearly see that the clothes and walls are also not finished rendering as they are all quite grainy. But yes, the skin is the most complex material in the scene and takes the longest to reach convergence.

 

[anne O'nymous]

The skin materials are identical. What you are seeing if the result of it failing to produce enough iterations for the skin to be completed.

Are you really sure ?
What you get is all elements looking the same, except the skin. What clearly express that the said skin need more iterations than the other materials to be fully proceeded. And this really point to its material ; probably the Translucency weight that is different, but it can be another value, it's just the most usually tweaked one.

And while I'm here again, there's something that bother me since my first answer: You need more than one hour to render the first scene with a RTX 3070 ? What the fucking fuck ?
Alright it's a big resolution, but yet I've seen more complex scenes rendered in less than 20 minutes with a RTX 20xx. Either Daz don't support the card yet, or there's something really wrong in your rendering configuration.

 

[slitherhence]

Here's the first scene limited to 230 iterations (took 10.5 minutes).

Spoiler: Intentionally hobbled render

You don't have permission to view the spoiler content. Log in or register now.

You can see the results are exactly the same. So the issue is that the iterations are taking longer in the second scene. So much longer that it makes it impossible to render the scene in a reasonable time. That's what I don't understand.

It's worth noting that my RTX _is_ bottle necked due to my CPU/MB being PCIe v3.0 (haven't been able to afford to upgrade it yet)... but so far I've only seen a small amount of difference from published benchmarks and my RTX 2070 regularly took twice as long to complete the same renders as my RTX 3070.

Also worth noting that the hair is strand based dforce hair... which may be a contributor to the render times being longer than you expect them to be. But it's the same hair in both scene and i haven't actually run the dforce sim for either scene cause I don't believe these scenes need it.

Spoiler: System Info

You don't have permission to view the spoiler content. Log in or register now.

Spoiler: Render Settings

You don't have permission to view the spoiler content. Log in or register now.

 

[mickydoo]

Turn off Render Quality Enabled and just manfully set it for 3000 iterations.
Put instancing optimization on auto.
Untick CPU in the advanced tab, unless you have a super dooper NASA cpu it will slow you down.

I have an 360ti, at a guess it would take me 10 minutes to render that scene.

 

[slitherhence]

Turn off Render Quality Enabled and just manfully set it for 3000 iterations.
Put instancing optimization on auto.
Untick CPU in the advanced tab, unless you have a super dooper NASA cpu it will slow you down.

How will setting it to max 3000 iterations cause iterations to take less time? If it won't... how will it reach 3000 iterations when it cuts off at ~230 iterations after 2 hours (you didn't say to disable the max time)? And assuming I disable the max time... do you really expect me to spend 26 hours on each render? That's not going to happen.

The instancing optimization setting has no meaningful effect on render times. And I already have CPU rendering disabled, always have.

I have an 360ti, at a guess it would take me 10 minutes to render that scene.

I very seriously doubt that. But here's the scene file, feel free to try...

You must be registered to see the links

. All assets should be available on f95zone.

 

[mickydoo]

How will setting it to max 3000 iterations cause iterations to take less time? If it won't... how will it reach 3000 iterations when it cuts off at ~230 iterations after 2 hours (you didn't say to disable the max time)? And assuming I disable the max time... do you really expect me to spend 26 hours on each render? That's not going to happen.

The instancing optimization setting has no meaningful effect on render times. And I already have CPU rendering disabled, always have.



I very seriously doubt that. But here's the scene file, feel free to try...

You must be registered to see the links

. All assets should be available on f95zone.

Ok 15 minutes, point being its not a very complicated scene, if its taking you an hour with 3070 you are doing something wrong, but if you seriously doubt that I can't help you.

 

[slitherhence]

Are you really sure ?

Yes, I have now manually compared every single property of both figures' skin material. They are identical. And my previous comment also demonstrates that, when limited to a very low number of iterations, the first scene behaves exactly the same in relation to the skin and other materials in the scene.

And, frankly, I'm losing patience with people's opinions about how complex they believe the scene to be. It is not relevant to the issue at hand. Any complexity present or lacking in one scene should be present or lacking in the other scene ergo the render times should be the same... yet they are not. That is the subject at hand. I don't think you intended to derail this thread but as that is now what is happening I need to ask you to drop it.

 

[probably_dave]

Yes, I have now manually compared every single property of both figures' skin material. They are identical. And my previous comment also demonstrates that, when limited to a very low number of iterations, the first scene behaves exactly the same in relation to the skin and other materials in the scene.

And, frankly, I'm losing patience with people's opinions about how complex they believe the scene to be. It is not relevant to the issue at hand. Any complexity present or lacking in one scene should be present or lacking in the other scene ergo the render times should be the same... yet they are not. That is the subject at hand. I don't think you intended to derail this thread but as that is now what is happening I need to ask you to drop it.

These type of problems are very annoying and can be caused by a few different things.

Starting with the basics, make sure it's rendering in GPU. I don't trust the 'Disable CPU' option so always check the log to ensure it's not automatically falling back to CPU (clicking the 'History>>" button on the render popup and ensuring it does not mention CPU anywhere). I'm assuming it isn't but this should always be the first thing that's confirmed.

Next would be lighting. Additional lighting in the scene would cause the slow down for each iteration. You mentioned it's the same scene so it's doubtful this would be the cause, but worth double checking, especially if your model has any emissive surfaces.

The next is to check, is usually something that is now present in the render window that wasn't present before that iRay is having difficulty with. This is likely your cause as you've only changed the pose.
Looking at your renders, the biggest difference is see is between the Lips and the Teeth. They appear silver in the second and glossy blue in the first. If these we're causing problems, the increase size of these in the second render could be exponentially increasing the render time. Try changing them to a basic shader to see if that help.

If not, you'll need to follow the standard trouble shooting process. Start with the scene that works fine, and adjust the model pose bit by bit until you encounter the problem. It's cumbersome to do, but it should be faster than the 26hrs it would take to render.

 

[anne O'nymous]

And, frankly, I'm losing patience with people's opinions about how complex they believe the scene to be. It is not relevant to the issue at hand.

I hesitated for a long time before answering you, wondering is someone who ask for help, then act as if he know better when the answer don't please him, effectively deserve to be helped.

So, the "not relevant to the issue at hand" way too high time needed to render the scene have an explanation, and this explanation can possibly be the reason why there's such difference in the time needed to render the two scenes.


[ mickydoo don't hesitate to correct me if I'm wrong, you know Daz way better than me]
What your stop at 230 iterations attempt show, is that everything in the scene is near to its final state in 10 minutes, except the skin. What mean that it's the skin that need the additional more than 1 hour needed to render the first scene.

Since the natural guess (a change in the material definition) isn't the cause, and since you provided a duf for one of the scenes, I decided to look closer. And I found, hmm, let's say "weird stuff" :

Spoiler

You don't have permission to view the spoiler content. Log in or register now.

In human language, it mean that you've only one character, but its skin materials are defined five times, and among them, four are definitions intended for 3Delight.
Being far to be an expert with Daz, I don't know if it's normal or not, for me it's just weird, but Daz is weird sometimes, so...


And this explain why Daz need more than one hour to render the scene. For everything except the skin, it's done fast, but unlike the rest, the skin isn't one material to render, but five materials to render then to blend together. This blending being the reason why it don't just need "five time more" times to render the scene.

Therefore, since you found no difference between the two scenes when looking in Daz, I wonder what you would found if you look directly into the duf files. My guess is that the second scene have the skin materials defined more times than the first one.

 

[DaCreat0r]

There are two ways a raytracing render can be slow:
1. Render needs many passes to reach desired convergence
2. Render passes themselves take too long

The first in indoor scenes is usually caused by too much differently coloured lighting or very poorly lit areas from my experience and obviously not the issue here.
The second is a little trickier to track down. If the passes themselves take long, that means that the system has to calculate more stuff per pass. This can be caused either by too many light sources, so there are just too many light rays bouncing around or by materials that scatter the light in ways that are difficult to calculate. Strange subsurface-shenanigans are the most likely issues. SSS can be hard on the GPU because the calculations needed to portray that effect are complex. That being said, the scene you presented is not a scene where that should be an issue. Either it's as anne O'nymous said and there is a bug with your material or the SSS make the caustic sampler go haywire. Try and turn that off. You only need the caustic sampler in scenes with caustics (e.g. glass that reflects on the floor, coffee mugs that reflect light onto the coffee, etc.) in it. It can kill render times otherwise.

The render times you write about are too high for your GPU if the settings you posted are the ones you always use. That is most likely an issue with the lighting (the lower half of your character is very noisy) and/or with the room itself. Try either using one of those indoor cams with a section plane attached to them or hide the walls that aren't visible in your render. That reduces the amount of bounces a light ray needs and thus speeds up times greatly.

 

[slitherhence]

I hesitated for a long time before answering you, wondering is someone who ask for help, then act as if he know better when the answer don't please him, effectively deserve to be helped.

And yet, I did know better. I, myself, said the skin was taking longer to render than other materials in the same scene multiple times. It was you that insisted the skin couldn't be taking longer than other materials yet was wrong. To be clear, it goes without saying that I am doing something wrong. It's just that using a different skin material in the two scenes isn't that "something wrong"... and the reasoning you used for why it had to be that didn't hold up given my much greater experience with these two specific scenes and even my very limited knowledge of DS. I've rendered both scenes probably 20 times trying to figure out what's different and I already knew the skin was rendering the same way in both scenes. I tried to gently explain to you this fact despite your dogged and condescending insistence to the contrary. And for that you claim I am potentially undeserving of help? Perhaps you now see the issue with your behavior. I'm willing to let it go if you do.

Since the natural guess (a change in the material definition) isn't the cause, and since you provided a duf for one of the scenes, I decided to look closer. And I found, hmm, let's say "weird stuff" :

Spoiler

You don't have permission to view the spoiler content. Log in or register now.

In human language, it mean that you've only one character, but its skin materials are defined five times, and among them, four are definitions intended for 3Delight.
Being far to be an expert with Daz, I don't know if it's normal or not, for me it's just weird, but Daz is weird sometimes, so...


And this explain why Daz need more than one hour to render the scene. For everything except the skin, it's done fast, but unlike the rest, the skin isn't one material to render, but five materials to render then to blend together. This blending being the reason why it don't just need "five time more" times to render the scene.

Therefore, since you found no difference between the two scenes when looking in Daz, I wonder what you would found if you look directly into the duf files. My guess is that the second scene have the skin materials defined more times than the first one.

This is more along the lines of what I suspected. The skin material is, unfortunately, a bit of a frankenstien's monster. I needed to use the textures/maps from one skin with properties from another in order to get rid of a visible texture seam between the tail and the torso. So the skin uses the textures/maps from CC Avandreal (also the source of the tail) while the rest is CC Circe.

I... don't know where the 3Delight or other properties come from. Possibly because of using Brestacular/Golden Palace? Both define 5 different zones for their grafts, each with their own material. They also blend the torso material into them via their material copy scripts. If you are looking at the body (torso) material on the genesis 8.1 figure itself then it really shouldn't be like that... all I did was change the textures (diffusion, specular, etc) used.

You must be registered to see the links

is the other scene's duf for comparison.

There are two ways a raytracing render can be slow:
1. Render needs many passes to reach desired convergence
2. Render passes themselves take too long

The first in indoor scenes is usually caused by too much differently coloured lighting or very poorly lit areas from my experience and obviously not the issue here.

The HDRI I'm using for outside the windows is actually blue tinted light. And lighting in the scene was something I worried about. It's less bright than what I've been using when I was experimenting with character portraits... if only a little. If you say it's bright enough I'll take your word for it.

The second is a little trickier to track down. If the passes themselves take long, that means that the system has to calculate more stuff per pass. This can be caused either by too many light sources, so there are just too many light rays bouncing around or by materials that scatter the light in ways that are difficult to calculate. Strange subsurface-shenanigans are the most likely issues. SSS can be hard on the GPU because the calculations needed to portray that effect are complex. That being said, the scene you presented is not a scene where that should be an issue. Either it's as anne O'nymous said and there is a bug with your material or the SSS make the caustic sampler go haywire. Try and turn that off. You only need the caustic sampler in scenes with caustics (e.g. glass that reflects on the floor, coffee mugs that reflect light onto the coffee, etc.) in it. It can kill render times otherwise.

Have tried without caustic sampler. It actually makes no appreciable difference for either scene's render times. I have also heard that it increases render times and can't explain why it seems to be having no effect here. The only reason I turned it on is the large windows right above the bed and the light coming in via those windows from the HDRI.

I haven't counted the light sources in the scene... but it would not surprise me if there was a dozen of them. Each of the computer screens in the back ground are emissive and there are several overhead lights. The set I am using is Space Station Living Quarters.

The render times you write about are too high for your GPU if the settings you posted are the ones you always use. That is most likely an issue with the lighting (the lower half of your character is very noisy) and/or with the room itself. Try either using one of those indoor cams with a section plane attached to them or hide the walls that aren't visible in your render. That reduces the amount of bounces a light ray needs and thus speeds up times greatly.

Section plane? Hmm. I've seen properties for that (in the camera properties?) but know nothing about it. I'll look into it.

As I said, the lighting is mostly overhead lighting. I rather do want the underside of the figure's body to be less well lit as that's how it would actually look given overhead lighting and a person laying in bed. I typically render at twice the resolution I plan to use and then resize it down to help smooth out inconsistencies while preserving detail. And I can deal with the remaining noise using nvidia's denoiser if I need to.

I can easily place a large white plane primitive behind the camera and disable the walls and such on that side. Can't leave it open cause of the HDRI.

That said. Everything you have brought up here is off topic. Right now my focus is on the difference between these two scenes. Everything you have brought up is identical between the two scenes and thus cannot explain why one scene renders in an hour and the other would take twenty six hours. PLEASE, stop going off topic. I'm going to have to start reporting people if this keeps up. And you, norgur, have at least been polite and respectful. But I have an issue that needs resolving before I pursue this other issue you guys keep insisting I have. Fine. I may have another issue. But it's not the one that this thread is about.

 

[mickydoo]

[ mickydoo don't hesitate to correct me if I'm wrong

Nah I'm out, I don't like typing for the sake of typing, so when I try and help and am seriously doubted and my post facepalmed after rendering 20,000 images myself by trying to get to core of the issue, there is nothing I can do. And, frankly, I'm losing patience with people's opinions on seriously doubting in general (and facepalming)

 

[slitherhence]

The next is to check, is usually something that is now present in the render window that wasn't present before that iRay is having difficulty with. This is likely your cause as you've only changed the pose.

Looking at your renders, the biggest difference is see is between the Lips and the Teeth. They appear silver in the second and glossy blue in the first. If these we're causing problems, the increase size of these in the second render could be exponentially increasing the render time. Try changing them to a basic shader to see if that help.

Yes, obviously some difference has been introduced. The lip and teeth material hasn't changed but it is definitely more exposed in the second scene. I had considered it before but, unfortunately, there are currently several other scenes in this set I haven't talked about where the mouth is closed but the issue remains. Still, I will try what you suggested.

If not, you'll need to follow the standard trouble shooting process. Start with the scene that works fine, and adjust the model pose bit by bit until you encounter the problem. It's cumbersome to do, but it should be faster than the 26hrs it would take to render.

:cries: I really don't want to. But its starting to look like I may have to.

 

[DaCreat0r]

That said. Everything you have brought up here is off topic. Right now my focus is on the difference between these two scenes. Everything you have brought up is identical between the two scenes and thus cannot explain why one scene renders in an hour and the other would take twenty six hours. PLEASE, stop going off topic. I'm going to have to start reporting people if this keeps up. And you, norgur, have at least been polite and respectful. But I have an issue that needs resolving before I pursue this other issue you guys keep insisting I have. Fine. I may have another issue. But it's not the one that this thread is about.

Actually none of it is. Everyone here as tried to point out what could be the issue. I have pointed out that besides the lighting an issue with the material could be the cause. Have you swapped materials yet?

If it is something that has become visible through the change in poses (which would be strange in that setting) then hide objects one by one and see what happens.

 

[anne O'nymous]

Everything you have brought up is identical between the two scenes and thus cannot explain why one scene renders in an hour and the other would take twenty six hours.

Have you tried to understand what he said, instead of just discarding it because you know better ?
As he implied, and obviously depending of the material, a simple change in the pose can lead to the light rays having to bounce way more, and therefore to the scene needing more time to be rendered. This is an unavoidable issue that we all encounter, but its impact can be (largely) minimized by cleaning and optimizing the scene.

Wanting to do too much is something that we all did. We all started by rendering few scenes with 100% realism ; keeping all the walls because they impact the lighting, putting too much lights because, like photographs, we know that it's the right lights that make a photo (and so a render) look prettier, and so on.
Doing this, we also learned that it's not how it should be done ; that the implied difference is too subtle to worth it with the kind of computers we have. We aren't rendering an animation movie and we don't have a farm of thousand computers dedicated to the rendering.
We all learned about it, and were trying to speed-up this learning process for you. It's apparently not the kind of answers you expected, but it's the kind of answers you needed ; the kind we all would have loved to have when we started : unless it's your job, rendering is before everything else a question of compromise.

PLEASE, stop going off topic.

Well, the problem is that, the more you talk about your issue, the more we are on topic.
As you said yourself, expliciting it a little further every time, the two scenes are identical except for the pose. Therefore, the issue you're encountering is due to the rendering configuration. And by this I don't limit to the value you entered but also, as norgur said, to the lack of optimization of the scene itself.
Be noted that this lack of optimization can also lead Daz to render everything relying only on the CPU, because the GPU RAM is too small. What would also have a strong impact on the time needed to render the scene. But I doubt that it's the case here, the scene is too small to not fit on the 8GB of your RTX 3070.
And by following the advice gave to you, you can also simply remove the issue you encounter. Not magically, but removing the unnecessary light rays bounce can lead to the two scenes being finally effectively totally identical, even in their rendering time. This also, we all encountered it when we started.

But well, the fact that you ask for help and immediately jump on your high horse because the answer you get don't please you make me join mickydoo. Good luck with your issue, and also with your game, since every scene with what is probably a main character will take 6 time too many times than effectively need to be rendered.

I'm going to have to start reporting people if this keeps up.

Please, do. I wonder how the moderators will react face to your obvious lack of respect for the people who try to read your mind and guess your whole configuration, this in order to provide the help you asked for.

 

[slitherhence]

But well, the fact that you ask for help and immediately jump on your high horse because the answer you get don't please you make me join mickydoo. Good luck with your issue, and also with your game, since every scene with what is probably a main character will take 6 time too many times than effectively need to be rendered.

That's not what happened. I asked for help, and then you pointed to a possible problem which I already knew was not correct from prior testing... and which has since been proven to not be correct. That's no big deal. Helping someone is a process. I, in fact, do not expect you to read my mind. Which is why I gently tried to explain to you that the issue you thought you were seeing wasn't actually there. It was then that _you_ got up on your high horse and began condescending to me insisting that it wasn't possible for the skin material to take longer than the other materials in the scene (something that even at my level I know isn't correct) and therefore the skin material in the second scene MUST be different from the first. I remained patient with you and explained again, in greater detail, why I had strong reason to doubt the skin materials being different between scenes. You responded with more condescending. So I stopped and did all the needed work to prove the skin materials were the same and succeeded in proving you wrong... and you responded by declaring that I wasn't deserving of help. And now you are engaging in historical revisionism pretending none of that ever happened.

I have been respectful throughout. Respect does not mean stroking your ego or letting you walk all over me. Respect means treating others as equals. Something I have done and you have not. I'm very conscious of the need for further optimization. But that is not the subject at hand. And you don't get to come into my thread and decide that the thread should be about something else. That is the very opposite of respectful.

Again I do not expect anyone to read my mind. Which is why I have fallen over myself to provide as much information as possible. But many of you have outright ignored the information provided when it didn't suit your agenda. For example, continuing to suggest that IRAY is resorting to CPU rendering despite me repeatedly pointing out that I have CPU rendering disabled and have confirmed that the logs show the GPU is being used. It's you people that have repeatedly rejected answers you didn't like.

I am extremely disheartened by the way you people have behaved. The egotism and now outright lies. I just don't know what else to say. It's clear this thread has been rendered unsalvagable. I'll have to make a new one. I'd say I'm starting over... but there's been no forward progress at all so... it's not really starting over is it?

 

[mickydoo]

I am extremely disheartened by the way you people have behaved. The egotism and now outright lies.

Says the person who doubted me and face palmed me. If you want to dob us all in to mods like a 6 year old go your hardest. No one knows the answer to your problem, hence we have to trouble shoot. If your renders are taking an hour it could be half the problem, who knows, well no one now.

 

[anne O'nymous]

It was then that _you_ got up on your high horse and began condescending to me insisting that it wasn't possible for the skin material to take longer than the other materials in the scene (something that even at my level I know isn't correct) and therefore the skin material in the second scene MUST be different from the first.

Let's see:

What you get is all elements looking the same, except the skin. What clearly express that the said skin need more iterations than the other materials to be fully proceeded. And this really point to its material ; probably the Translucency weight that is different, but it can be another value, it's just the most usually tweaked one.

Not only there's nothing along the line of "it's not possible for the skin material to take longer" (it's even the opposite), but there's also no "must" (and even less a BCP 14 "MUST"), but a "point to" that is far to have this imperative (definitive / BCP 14) tone.

[...] and you responded by declaring that I wasn't deserving of help.

False once again:

I hesitated for a long time before answering you, wondering is someone who ask for help, then act as if he know better when the answer don't please him, effectively deserve to be helped.

I didn't declared, I wondered aloud, and I totally assume it. It was an intentional hint that you clearly missed.
And there's really no way for someone to seriously take this as an affirmation, since while writing this, I still took the time to look in detail at your 270 KB scene, and to wrote tenth of lines, still trying to help you. What I obviously wouldn't have done if effectively I thought that you didn't deserved to be helped.

And now you are engaging in historical revisionism pretending none of that ever happened.

At no time there's, in my previous messages on this thread, something that can be interpreted like that. As for this message, I just correct your obvious lies.
Note that, unlike you, I decided to name them "lies", in order to not insult my interlocutor by talking about "historical revisionism".

Respect means treating others as equals. Something I have done and you have not.

You didn't :

And, frankly, I'm losing patience with people's opinions about how complex they believe the scene to be. It is not relevant to the issue at hand. Any complexity present or lacking in one scene should be present or lacking in the other scene ergo the render times should be the same... yet they are not. That is the subject at hand. I don't think you intended to derail this thread but as that is now what is happening I need to ask you to drop it.

You're clearly assuming that anyone else than me who tried to help you didn't cared to read your thread opening, your other messages, and to think by themselves. Instead, you present them like sheep that saw me talk about the lack of optimization of your scene, then decided to blindly follow.
In what universe is this a respectful behavior ?

I'm very conscious of the need for further optimization. But that is not the subject at hand. And you don't get to come into my thread and decide that the thread should be about something else. That is the very opposite of respectful.

And once again, this obvious lack of respect you have for anyone who wrote on this thread. As if they pointed the lack of optimization just because I did, and not because, for the experienced persons they are, it really seem to be the effective cause of your problem.

For example, continuing to suggest that IRAY is resorting to CPU rendering despite me repeatedly pointing out that I have CPU rendering disabled and have confirmed that the logs show the GPU is being used.

I only talked about the CPU once:

Be noted that this lack of optimization can also lead Daz to render everything relying only on the CPU, because the GPU RAM is too small. What would also have a strong impact on the time needed to render the scene. But I doubt that it's the case here, the scene is too small to not fit on the 8GB of your RTX 3070.

But it wasn't to "[continue] to suggest that IRay is resorting to CPU rendering", but to explicitly say the opposite.

And I was just talking about a well know fact. If the whole scene don't fit in the GPU RAM, Daz will not magically generate more RAM, it will pass over its configuration and use the CPU, dot.

It's you people that have repeatedly rejected answers you didn't like.

Say the person who've, so far, rejected all the answer, stating that "it's not the problem".

I've been on this Earth since near to 50 years, and I past near to 35 years on forums (BBS before internet). I've seen thousands of people asking for help, just to discover that the problem wasn't at all what they thought it was ; it even happened to me, as for anyone else.
But no, you aren't one of those. You admit yourself that you don't have a single clue about what can be your issue, but there's one thing you're absolutely, totally, and definitively, sure about, the lack of optimization cannot be the issue you're facing. And you're sure about this even after having been explained, by more than one person, why and how it can effectively be at least part of the issue you're facing.

But you must be right, we are the ones rejecting the answer we don't like. It make no sense since we also are the ones answering, but you're surely right anyway.