Hey y'all just ignore me here. I've had this hateful flu for the past 3 days and have had maybe 5 hours of sleep since it began, if that. After re-reading my own last few posts, they don't even make sense to me. I'm not sure exactly what I'm saying or thinking, I think I've gone delirious. ;-)

Quote - But aside from that, if the renderer is expanding an image size, isn't it getting stretched slightly?

Mike - didn't think you were head-butting me :)

But to answer that question, I think the answer is no it doesn't. I think it gets expanded, not re-sized. Meaning, the extra unused pixels are filled with nothing in particular. They just take up space. The UV coordinates are just mapped to 0 to 3999 instead of 0 to 4095.

Quote -

But to answer that question, I think the answer is no it doesn't. I think it gets expanded, not re-sized. Meaning, the extra unused pixels are filled with nothing in particular. They just take up space. The UV coordinates are just mapped to 0 to 3999 instead of 0 to 4095.

Ah, I see, thank you for the 'splain. Makes sense. I should have seen that.

It's a pity we did not carry on with the OP's subject. I think it could have led to a useful discussion covering such things as good basic lighting, and perhaps some techniques to improve the appearance of scene props and stuff that come with poor shaders or textures.

Oh well... he missed an opportunity there. We all did.

As an aside, that (very informative) article that bb linked to started off by saying that the Shader Rate is our number one image quality knob. I tend to disagree with that. I think that our number one image quality control is render pixel dimensions. I did a test on Stonemason's Streets of the Med, rendering at 1200x1200 with a shading rate of 0.25, and then at 2400x2400 with a shading rate of 1.0. I did not time the renders exactly, but they took about the same time. I then reduced the larger image in Photoshop Elements to 1200x1200 (using bicubic resampling) and did a close up comparison. The larger render was far and away the better quality image, particularly in respect of anti-aliasing and more accurate reproduction of finer details such as the ivy leaves.

There are probably clear technical reasons why that should be the case, but this empirical evidence is more than enough for me to start doing larger renders in preference to reducing the shading rate.

MikeJ... I hope your bug eases off tonight, so you can get some good sleep and enjoy the weekend.

Quote -
It's a pity we did not carry on with the OP's subject. I think it could have led to a useful discussion covering such things as good basic lighting, and perhaps some techniques to improve the appearance of scene props and stuff that come with poor shaders or textures.

If it helps any, I'm going to be launching a thread soon showing some of the really bad and weird things about Poser rendering I've come across, and hopefully it will become a thread with some good workaround information or some good do's and dont's. I just have to gather up some screenshots and such, but watch for that in the next few days.

Thank you for the well-wishes. :-)

Good! And I'm preparing a lighting tutorial. Couple people have asked.

Quote - I agree, it doesn't really make sense that the render settings would be an issue.

Doesn't sound like the render engine is the issue.
It's the content, different shader setups can look different under different lighting.... but I don't see anything exotic about that, it's just the nature of 3Ding, textures and lighting need to work together, and there's no 'single standard setup' to make everything work with.

I think that's what's giving the OP grief, he wants all things in a scene to work the same under a light set that he picks, without having to make tweaks to shaders, and is peeved that vendors aren't doing it for him.. sees it as a flaw in the product.

(Sorry, I didn't read the whole thread, only made it as far as the post I replied to)

Quote - I never will understand why only in Poser do people make "off size" image maps.
The whole professional 3D world works in powers of two for textures, and there's an actual physical reason for that, but nooooo... not for Poser. Poser has to be different, I suppose.

IRRC, there used to be a limitation, if you want to make your product MAC compatible, the texture has to be no bigger then 4000x4000. Some sort of Poser+MAC hardware limitation. I actually don't know if that limitation still exists, but a lot of content makers are in the habit of making 4000x4000 textures, stemming from that limitation.

Quote - It's a pity we did not carry on with the OP's subject. I think it could have led to a useful discussion covering such things as good basic lighting, and perhaps some techniques to improve the appearance of scene props and stuff that come with poor shaders or textures.

If you do a bit of searching for posts and threads by Bagginsbill, you'll find heaps of wonderful information on this subject.

:: If you mean me, Conniekat8, I've probably read nearly all of them!

I was thinking of others, not myself. Unfortunately the very heaps of information that you refer to can be very confusing and too much to take in for those who do not even know what they need to know, or what questions to ask.

If only for the sakes of such Poser users, I welcome the suggestions by MikeJ and bb for fresh introductions to these topics, and I hope they can be kept on track and not get too deep, or diverge too much from their purpose.

It's tough taking a topic that started off on such a negative note into something positive.... You know, something that starts with, everyone else is doing it to me to mess me up, my mind is already made up

rather then for example a thread that starts off with, why is this happening, how do I deal with it, or I'm lost, please help...

If you really want to do an intro for others, just start a new thread, one that starts off on a more positive note and run with it....

Agreed. I'm expecting that's how MikeJ and bb will handle it.
I don't know enough to provide any such help myself!

Quote -

As an aside, that (very informative) article that bb linked to started off by saying that the Shader Rate is our number one image quality knob. I tend to disagree with that. I think that our number one image quality control is render pixel dimensions. I did a test on Stonemason's Streets of the Med, rendering at 1200x1200 with a shading rate of 0.25, and then at 2400x2400 with a shading rate of 1.0. I did not time the renders exactly, but they took about the same time. I then reduced the larger image in Photoshop Elements to 1200x1200 (using bicubic resampling) and did a close up comparison. The larger render was far and away the better quality image, particularly in respect of anti-aliasing and more accurate reproduction of finer details such as the ivy leaves.

There are probably clear technical reasons why that should be the case, but this empirical evidence is more than enough for me to start doing larger renders in preference to reducing the shading rate.

.

i tryed now this.

i rendered an iamge 600x400. shading rate 1
then i rendered the same image 1200x800. so one time bigger. then in photoshop i made the bigger render 600x400. and when i compare the two renders the second one is better quality.

hmmmm  i will need to do more tests now.  lower shading rate is better. but how better.
i almost always need to use blur after i render because the render from poser is super-sharp. to sharp.

You changed the shading rate by a factor of 4. Good.

You guys didn't mention the other two factors that are altered when you double the render size, which means 4x the area. Since you didn't mention them, I assume you didn't think about them, and they matter.

1. Did you change the pixel samples to 4x on the smaller render? If not, you were sampling the double size image at 4 times the samplling rate of the smaller image. To do a comparison, you should use something like pixel samples = 16 on the small one and pixel samples = 4 on the large one.

2) Did you have post filtering turned off or minimized? If it is on, you cannot compare apples to apples, since the post filter is applied to the rendered pixels, and on the double size render this becomes a much more fine-grained filter.

I just did a test taking these into account and I can see no difference.

i have filter set to 1.

the question now is if i make a 1200x1200 render or a similar size do i need shading rate 0,2? 

bb, thanks for your feedback.

This wasn't in any way intended as a scientific test.... I just wanted to do a direct comparison between the effectiveness of increasing the render size and reducing the shading rate, without changing anything else. 4x the area is why I tried a direct comparison with 0.25x the shading rate. Yes, I did think about it, even though I didn't mention it!

My result showed quite clearly that increasing the render size by a factor of 4 improved quality more effectively than decreasing the shading rate by a factor of 4, and also had little impact on render time. (I'd have to go back and time the renders to be absolutely sure about that claim, though).

This is a single test result and may be a generalisation, but it's one I'm willing to go along with for the moment.

Of course, we would combine the two controls, with the other settings as well, to give even better quality, at the expense of extra render time.

(edit) I was not using any postfilter.

OT: 
I had a terminal Poser Pro crash last night, and in the end I had to re-install the program. I'm still sorting out the SRs and the scripts and V4 stuff that were installed in my main Runtime. At least it's grey and raining outside, so I don't feel too bad about sitting here at my desk today!

Well there was no need to do the test, because if you kept the same shading rate then of course the double size render was better. I could have told you that without even trying. Why wouldn't it be? Since you effectively increased the sampling rate by a factor of 4, because you didn't adjust it for the new size, this should be a much better, more detailed render, with less aliasing.

Rendered with

pixel samples = 16
shading rate = .25

Pixel samples = 4
Shading rate = 1

It looks about the same, maybe a little better than the 400x400 render. (You need to blow these up big to compare.)

However, I will say this one took 10 seconds to render, versus the smaller render took 48 seconds!

(Edited for saying the opposite of what I meant to say)

Excuse me, bb, but you are not reading my posts properly. I did not keep the same shading rate in the smaller render, I reduced it by a factor of 4, the same as the increase in size for the other render.

And there was a need to do the test. You know lots of things that I can only find out by experiment. You are the one who always says "show me"!

:O)

''. Why wouldn't it be?''

i didnt know this. i only lern from you. i maybe sometimes read a paper that i find on the internet. but thats it. the biggest problem with me is that sometimes i dont understand the ''hard '' words. to  all of you maybe simple words. to me (from europe) not. so its easier for me when i see  examples on pictures. when i see i understand it easier.

well now i know so cool.

Quote - Excuse me, bb, but you are not reading my posts properly. I did not keep the same shading rate in the smaller render, I reduced it by a factor of 4, the same as the increase in size for the other render.

And there was a need to do the test. You know lots of things that I can only find out by experiment. You are the one who always says "show me"!

:O)

Ummm, seems the other way around. You have not once used the words I'm using.

You adjusted only one parameter - shading rate. You did not change the other parameter - pixel samples.

When you double the render size, you are effectively quadrupling BOTH parameters. You only adjusted one of them. The result was you were not doing equivalent values.

(EDIT: Again, I got my words mixed up. I'm talking about pixel samples.)

I chose the pattern because I wanted something that had various sizes of horizontal lines, vertical lines, diagonal lines, and lines that were in between, very close to but not exactly horizontal or vertical.

When you save the material, remove the .txt at the end and put it in your materials folder.

bb, please.... I repeat the words that you appear to have overlooked:

"I just wanted to do a direct comparison between the effectiveness of increasing the render size and reducing the shading rate, without changing anything else."

Which is exactly what I did. I did read your posts very carefully, as I always do. But I did not want to change the pixel samples, since that is an additional factor that I did not want to include in my comparison.

If your second test had used the same pixel samples (16) it would presumably have taken around the same time as the first one to render, and would presumably have given better quality. That would have been repeating the comparison I intended.

OK, so if I understand correctly, your decision to keep the pixel samples the same was a conscious decision, not an inadvertant omission from the test.

So my response is the same but with a different spin on it.

I'm claiming that your test demonstrated only the obvious. Given two renders where one has 4 times the effective sampling rate as the other, the one with the higher effective sampling rate will look better.

To me, test or no test, it's like you announced the "discovery" that sample rate can improve image quality. We knew that already.

The reason I'm pointing this out so annoyingly is I don't want people to start trying to improve image quality with 4K by 4K renders for no reason.

PS: I still love you. :)

so when i do a big render i dont need low shading rate? :) 

or do i he he

(edit) I still love you too! but i'm not changing what I wrote below! It's from the heart.

Correct, bb, it was a conscious decision (at last!)

Increasing render size obviously improves quality. Decreasing shading rate obviously improves quality. I was simply trying to compare the two methods of improvement side by side without any other factors involved.

But you are wrong, "WE" did not know the results already. YOU knew that already, and perhaps some others here. I, and many others, did not know. For me the comparison was a discovery, and there is nothing wrong with letting others know about it. Your sarcasm about me making an announcement of a "discovery" that sample rate can improve image quality (which is not what I announced at all) is really unbecoming.

Perhaps you would prefer it if I kept my little nonsense experiments to myself in future, since YOU already know the results of them?

I'm shaking my head in sheer disbelief here.
It's such a shame that genius so often comes at the expense of graciousness.

I'm sorry for being so outspoken and possibly annoying you, but really, bb, you deserve to be told. Please think on it.

Hehe - this is getting complicated.

No doubt, increasing the effective shading rate and/or the effective sampling rate makes an improvement. You can accomplish these either by making a bigger render then reducing, or by changing the values while keeping the same render size, or by doing both.

The choice of which technique to use should be made on the basis of all factors. The factors, as much as occurs to me at the moment, are:

1) Image quality
2) Render time
3) Memory use

As usual, these things are inter-related and not always linear. Meaning, if I double the effective shading rate, I will probably alter the render time, but it may be more or less than double. If I double the memory use, but it is still under my total physical RAM, it will go fast, but if I happen to go even slightly over my total RAM space, and I start swapping, then render time will go up 100x or more. I've seen renders that triggered swapping take days.

I do not know all the precise points at which RAM is exhausted, but I have heard it can happen just by making a larger render. In such cases, it may be better to change the parameters instead of changing render size. On the other hand, I've had at least one test where changing the parameter was much slower than changing the render size, even though the effective shading rate was the same.

Quote - I'm shaking my head in sheer disbelief here.
It's such a shame that genius so often comes at the expense of graciousness.

I'm sorry for being so outspoken and possibly annoying you, but really, bb, you deserve to be told. Please think on it.

Hey, I wasn't annoyed! <Kiss, kiss>

I worried about some noob thinking to improve image quality by going to a 4K render and getting upset that Poser crashes with out of memory.

The whole point of the REYES renderer, and all this shading and sampling within a bucket, is to be able to process large amounts of scene data with limited physical memory.

I just did another few trials, trying to get rid of the moire pattern in my test sample.

I cannot do it by rendering 400x400 with Pixel Samples = 16 and Shading Rate = .0625. (Which is 1/4 of .25) Image quality did not improve, while render time was much worse.

But I can do it by making the render 800x800 with Pixel Samples = 4 and Shading Rate = .25, then reducing to 400x400. Render time did not change much.

So, Image Quality is much better and render time is much better. Those are the two biggest factors we care about, and since BOTH improved simultaneously, I declare the following:

The better technique for desired image quality on this test pattern with a single polygon is to render big, instead of changing parameters. For reasons unknown, the effective shading and sample rates do not come out the same.

You all can thank IsaoShi. I'm an ass.

Pixel Samples: 64
Shading Rate .015625 (1/64)

The results are the same. There is a fundamental limit here in 400x400 pixels that cannot be improved upon.  These fine lines that are nearly vertical or horizontal will always show a moire pattern - they cannot be eliminated.

However, the bad moire in the upper right corner is gone and it was already gone at 800x800 which was a very fast render. You might conclude as a rule of thumb that rendering at 2x the intended resolution will produce very close to the best possible image quality. The incremental gains of a 4x render (at least for this particular test case) are not worthy.

interesting.

well 400x400 is used only for test renders right? . we  always render bigger for a final rende.r

400x400 isn't for anything in particular.

Imagine my test pattern is actually a painting hanging on a wall in the background of an otherwise large render. It may end up being roughly only 400 pixels within the scene, even if the overal render is 1280x1024. Now imagine that you look at that painting, and you notice that the picture frame looks pixellated because it is being viewed slightly off-center, so that the edge of the frame is not quite vertical or horizontal. How would you fix that? Before today, I might have suggested you improve the shading rate and the sampling rate, but it seems there are fundmanental limits to the value produced by these, where image quality doesn't get any better, but render time gets much worse than 4x. I see now the possibility that you should render double size then reduce in post, and the image quality will go up a lot, while the render time may not go up 4x.

ahaaaaaaaa.

dumb me. of course. i was only thinking about the objects that are infront of the camera. of course. if its in the back it will be small .

Quote -
I see now the possibility that you should render double size then reduce in post, and the image quality will go up a lot, while the render time may not go up 4x.

Used to have to do that with Vue too, to combat its tendency to produce grainy renders no matter how high the AA was set to.

Quote - Agreed. I'm expecting that's how MikeJ and bb will handle it.
I don't know enough to provide any such help myself!

Well, don't expect much out of me other than problems. Bagginsbill is the Poser expert, I'm just the grumpy user with alot of questions. ;-)

I pretty much use Lightwave these days mostly. I've re-rigged my favorite Poser people in LW, as opposed to using the export plugins, because Ligthwave offers just sooooo much more than Poser, from rigging to animating and rendering and everything else.

However, that's ALOT of work, and LW doesn't make it easy, considering its overall poor OpenGL performance, which Poser outshines by several degrees of magnitude. Plus, the figure meshes need a certain amount of tweaking to rig well in LW.

So when I want to do Poser stills, I'd prefer to use Poser, simply  because it's easier and quicker. The problem I'm having now is, I know Poser pro can render alot better than I can get it to do, but I expect more out of it and hit these problems which baffle me. I'm sure there are plenty of others who experience that too.

And not just problems. I'm convinced alot of it is bugs, but am not convinced they can't be worked around. When I get my screenshots and renders together, hopefully some of this stuff can be dealt with.

Quote - Good! And I'm preparing a lighting tutorial. Couple people have asked.

how to setup lights? or what settings to use?

or just a general lighting tutorial?