Re: Mental Ray render settings was "Re: shadow flicker"

["Steven Caron" <carons(at)gmail.com>]

its all good...

thanks Tim.

On 6/6/06, André Adam <a_adam(at)49games.de> wrote:

Hmm, no idea if I just missed the point, sorry in case I did... :)

Tim Leydecker wrote:

>> Ok, this takes us absolutely nowhere, either you don't want to
>> understand that both of us talk about the very same thing or my
>> english simply is too bad to make people understand what I'm talking
>> about. Anyhow, I'm off this thread now, there's everything said and
>> repeated at least once.
>
>
> Butting into this thread for being an expert at misreading the
> intended tonality of a post and also being gifted with a most
> natural talent to stir things up in a somewhat disturbing way:
>
>>> "...Mitchell filtering, which in most setups needs higher radii to
>>> create nice-looking imagary, many more samples are taken into
>>> consideration for a single pixel's colour value...."
>>>
>>> indeed you did...
>>
>
> I´d say Steve just wanted to admit he had missed that bit in your
> previous post and even had the balls to do it in public. One could
> find a trace of sarcasm in it if it where written in german but I think
> it isn´t meant to be taken that way - as it´s not written in german...
>
> "Ja, und?"
>
> "Oder was?"
>
> Cheers
>
> tim
>
>
> ----- Original Message ----- From: "André Adam" < a_adam(at)49games.de>
> To: <XSI(at)Softimage.COM>
> Sent: Tuesday, June 06, 2006 10:18 AM
> Subject: Re: Mental Ray render settings was "Re: shadow flicker"
>
>
>> Ok, this takes us absolutely nowhere, either you don't want to
>> understand that both of us talk about the very same thing or my
>> english simply is too bad to make people understand what I'm talking
>> about. Anyhow, I'm off this thread now, there's everything said and
>> repeated at least once.
>>
>>
>> Steven Caron wrote:
>>
>>> "...Mitchell filtering, which in most setups needs higher radii to
>>> create nice-looking imagary, many more samples are taken into
>>> consideration for a single pixel's colour value...."
>>>
>>> indeed you did...
>>>
>>>
>>>
>>> On 6/5/06, *André Adam* <a_adam(at)49games.de
>>> <mailto:a_adam(at)49games.de>> wrote:
>>>
>>>     "filtering doesn't decide how many samples are taken..."
>>>
>>>     Hey Steven, that's exactly what I've said, reread my mails.
>>>     However, the
>>>     filter size *does* decide on how many samples will be combined
>>> for a
>>>     single pixel *after* all rays are shot by creating an ellipse
>>>     around the
>>>     center of that pixel and collecting all the samples within the
>>>     given x/y
>>>     radii.
>>>     Your observations regarding the eye-rays are interesting,
>>> though. We
>>>     seem to have a few extra rays in there, but I see the basic idea of
>>>     samples = pixels with a sampling value of 0 still being valid,
>>>     don't you
>>>     think? At least I can live with a variance of less than ten
>>>     percent... ;)
>>>
>>>
>>>
>>>     Steven Caron wrote:
>>>
>>>     > /deep breath!
>>>     >
>>>     > /filtering doesn't decide how many samples are taken...
>>>     >
>>>     >
>>>
>>> file:///C:/Softimage/XSI_5.0/Doc/mental_ray/manual/node76.html#INDEX188
>>>     >
>>>     > "...The *filter* statement specifies how multiple samples are
>>> to be
>>>     > combined into a single pixel value..."
>>>     >
>>>     > and if you just take look at jitter thats only two definitions
>>> down
>>>     > from filter...
>>>     >
>>>     > "...Without *jittering*, samples are taken at the corners of
>>>     pixels or
>>>     > subpixels..."
>>>     >
>>>     > and if you turn jitter on. with AA min 0 and max 0... mental ray
>>>     > kindly turns jitter off :) why? Because it will introduce
>>> artifacts
>>>     > with that low of a sampling rate...
>>>     >
>>>     > //INFO : RC   0.2  warn   082008: jittering disabled because max
>>>     > samples < 1
>>>     >
>>>     > So according to you a 100 x 100 render would produce 10000
>>>     samples? (
>>>     > eye rays ) Lets test this!
>>>     >
>>>     > I setup a format of... 100 x 100. AA min of 0 and max of 0.
>>> Jitter
>>>     > OFF. Threshold rgba to 1.0 and filter type box 1 1, but threshold
>>>     > doesn't matter because of min and max samples being the same and
>>>     > filter doesn't matter because it only averages samples already
>>>     taken.
>>>     >
>>>     > turn progress on...
>>>     > //INFO : RC   0.2  info : rendering statistics
>>>     > //INFO : RC   0.2  info :   type
>>> number   per
>>>     > eye ray
>>>     > //INFO : RC   0.2  info :   eye rays
>>>     > 10816           1.00
>>>     >
>>>     > hmmm interesting!
>>>     >
>>>     > I am going to throw another variable in here... task size or tile
>>>     > size. this option will really increase or decrease your
>>> samples...
>>>     > smaller task size is going to give you more samples.. larger is
>>>     going
>>>     > to give you less. Here is a quick formula for predicting
>>> samples. I
>>>     > say predicting, because there is no way to know everything
>>> they are
>>>     > doing internally, short of being a mental images developer.
>>>     >
>>>     > task size + 1 * task size + 1 = samples per tile...
>>>     >
>>>     > To test this make a 32 x 32 resolution AA min 0 max 0 and turn
>>> your
>>>     > "tile size" to 32x32. You can leave automatic in cause most
>>> likely
>>>     > mental ray will choose this, but just in case force it. you
>>> know so
>>>     > you feel like your in control. You will get...
>>>     >
>>>     > 32 + 1 * 32 + 1 = 1089
>>>     >
>>>     > have fun.. do some more tests...
>>>     >
>>>     > André Adam wrote:
>>>     >
>>>     >> To add to this, the position sampled is only important for the
>>>     >> filtering algorithm that collects samples using a specific
>>> radius
>>>     >> around the center of the pixel. With an aa setting of 0 and a
>>> box
>>>     >> filter with a size of 1, indeed four samples will be
>>> considered to
>>>     >> retrieve the final colour value for this pixel. However, hardly
>>>     >> anyone uses a box 1/1 filter. When switching to the more popular
>>>     >> Mitchell filtering, which in most setups needs higher radii to
>>>     create
>>>     >> nice-looking imagary, many more samples are taken into
>>>     consideration
>>>     >> for a single pixel's colour value. However, the sample
>>> density shot
>>>     >> by the renderer *does not* increase in this scenario, with aa
>>> 0 it
>>>     >> stays one sample shot per pixel.
>>>     >>
>>>     >>
>>>     >> André Adam wrote:
>>>     >>
>>>     >>> Nope, with aa 0 you get exactly as many samples as you have
>>> pixels
>>>     >>> for the whole image. It definately is one sample per pixel. Not
>>>     >>> centered, but that doesn't make much of a difference, the
>>> spacing
>>>     >>> between the samples stays one pixel regardless of the position
>>>     >>> sampled. It's just about offsetting the sampling raster.
>>> Which btw
>>>     >>> gets randomised anyways when enabling the sample jitter option.
>>>     >>>
>>>     >>> Cheers!
>>>     >>>
>>>     >>>    -André
>>>     >>>
>>>     >>>
>>>     >>> Steven Caron wrote:
>>>     >>>
>>>     >>>> "...and one sample per pixel (aa setting 0)..."
>>>     >>>>
>>>     >>>> just like bernard and kim said about threshold being
>>>     misunderstood
>>>     >>>> this too is misunderstood.
>>>     >>>>
>>>     >>>> mental ray samples at the corners of a pixel. meaning 4 per
>>>     pixel
>>>     >>>> not 1. but not always 4 new samples because mental ray can
>>> share
>>>     >>>> samples with bordering pixels.
>>>     >>>>
>>>     >>>> *bracing myself now*
>>>     >>>>
>>>     >>>> steven
>>>     >>>>
>>>     >>>> On 6/2/06, *André Adam* < a_adam(at)49games.de
>>>     <mailto:a_adam(at)49games.de>
>>>     >>>> <mailto: a_adam(at)49games.de <mailto:a_adam(at)49games.de>>> wrote:
>>>     >>>>
>>>     >>>>     A really nice example covering this problem is a
>>>     hemisphere mapped
>>>     >>>>     with
>>>     >>>>     a clear night sky showing lots of small stars. If the
>>> min aa
>>>     >>>>     setting is
>>>     >>>>     too broad (and one sample per pixel (aa setting 0)
>>> might very
>>>     >>>> well be
>>>     >>>>     too broad), the rays will likely completely miss out
>>> certain
>>>     >>>> stars on
>>>     >>>>     certain frames which leads to the stars wildly
>>> flickering in
>>>     >>>>     animation...
>>>     >>>>
>>>     >>>>
>>>     >>>>     kim aldis wrote:
>>>     >>>>
>>>     >>>>     >I'm not trying to put you down here Bernard but there are
>>>     >>>>     situations where
>>>     >>>>     >settings like this aren't good and you should have some
>>>     idea of
>>>     >>>>     what sort of
>>>     >>>>     >image you're rendering before making decisions about aa
>>>     settings.
>>>     >>>>     In some
>>>     >>>>     >shots, messing with threshold won't make any difference
>>>     at all,
>>>     >>>>     raising max
>>>     >>>>     >settings will make little difference unless you bring up
>>>     the min
>>>     >>>>     settings.
>>>     >>>>     >You're right, threshold is largely misunderstood. Let's
>>>     try and
>>>     >>>>     shed some
>>>     >>>>     >light:-
>>>     >>>>     >
>>>     >>>>     >here's something you can try. Get a cube, unit 1. scale
>>>     it so
>>>     >>>>     it's really
>>>     >>>>     >long and really thin - 0.02, 0.02, 35.
>>>     >>>>     >now duplicate it 20 or 30 times and space each cube
>>> one unit
>>>     >>>>     apart in x. Now
>>>     >>>>     >shift them all back away from the camera so you get a
>>> good
>>>     >>>>     convergance. Lots
>>>     >>>>     >of thin, parallel lines in the distance, all converging
>>>     to the
>>>     >>>>     vanishing
>>>     >>>>     >point is what you're looking for.
>>>     >>>>     >
>>>     >>>>     >Now mess with the aa settings and pay close attention
>>> to the
>>>     >>>>     quality where
>>>     >>>>     >the cubes are nearly converging. Try 0,3 and mess with
>>> the
>>>     >>>>     threshold. Not a
>>>     >>>>     >lot of difference, is there, even if you set the
>>>     threshold really
>>>     >>>>     as low as
>>>     >>>>     >you can get it. In fact, changing the threshold varies
>>>     not a lot
>>>     >>>>     at all.
>>>     >>>>     >
>>>     >>>>     >now compare 0,2 with 0,3. Not much better. And I'll bet
>>>     if you
>>>     >>>>     try 1,2 it'll
>>>     >>>>     >be way better than 0,3. Again, regardless of the
>>> threshold.
>>>     >>>>     >
>>>     >>>>     >The point here is that there's a real danger of getting
>>>     it wrong
>>>     >>>>     if you
>>>     >>>>     >start generalising aa settings and don't take into
>>>     account the
>>>     >>>>     kind of image
>>>     >>>>     >you're rendering, as Helen is finding out, since
>>>     sampling grain
>>>     >>>>     is fine
>>>     >>>>     >detail. I just tried something here. Pushing aliaising
>>>     to 1,2
>>>     >>>>     with default
>>>     >>>>     >threshold gave me better improvement than doubling the
>>>     >>>> sampling size.
>>>     >>>>     >lowering the threshold improved things a bit but nothing
>>>     worked
>>>     >>>>     better - nor
>>>     >>>>     >was faster - than raising the min setting, bringing it
>>>     closer to
>>>     >>>>     the max.
>>>     >>>>     >
>>>     >>>>     >If you really want to see how well - or otherwise - aa is
>>>     >>>>     working, turn on
>>>     >>>>     >'view sampling' in the region diagnostic tab. It's most
>>>     >>>> revealing.
>>>     >>>>     >
>>>     >>>>     >-----Original Message-----
>>>     >>>>     >From: owner-xsi(at)Softimage.COM
>>>     <mailto:owner-xsi(at)Softimage.COM> <mailto:owner-xsi(at)Softimage.COM
>>>     <mailto: owner-xsi(at)Softimage.COM>>
>>>     >>>>     [mailto: owner-xsi(at)Softimage.COM
>>>     <mailto: owner-xsi(at)Softimage.COM> <mailto:owner-xsi(at)Softimage.COM
>>>     <mailto: owner-xsi(at)Softimage.COM>>]
>>>     >>>>     On Behalf Of
>>>     >>>>     >Bernard Lebel
>>>     >>>>     >Sent: 01 June 2006 16:23
>>>     >>>>     >To: XSI(at)Softimage.COM <mailto:XSI(at)Softimage.COM>
>>>     <mailto: XSI(at)Softimage.COM <mailto:XSI(at)Softimage.COM>>
>>>     >>>>     >Subject: Re: shadow flicker
>>>     >>>>     >
>>>     >>>>     >I would definitely use a min of 0 and a max of 3.
>>>     >>>>     >
>>>     >>>>     >Also, what is generally ignored or misunderstood is the
>>>     treshold.
>>>     >>>>     The
>>>     >>>>     >treshold color is a control for how sampling is done.
>>> Lower
>>>     >>>>     values means
>>>     >>>>     >that sampling occurs with lower contrasts, in other
>>>     words, the
>>>     >>>>     level of
>>>     >>>>     >sampling have more chances of getting closer to the max
>>>     value.
>>>     >>>>     Higher
>>>     >>>>     >treshold means a higher contrasts is tolerated before
>>>     requiring
>>>     >>>>     another
>>>     >>>>     >level of sampling, so less samples may be takne.
>>>     >>>>     >
>>>     >>>>     >I never work with treshold higher than 0.05, and had to
>>>     go as
>>>     >>>> low as
>>>     >>>>     >0.005 on few occasions. Note that at that point, if the
>>>     image is
>>>     >>>>     detailed,
>>>     >>>>     >you should consider ignoring adaptive sampling
>>>     altogether and use
>>>     >>>>     same min
>>>     >>>>     >and max values. If the image has significant amounts of
>>>     constant
>>>     >>>>     colors
>>>     >>>>     >(like shadow passes, black backgrounds in the likes),
>>> then
>>>     >>>> adapative
>>>     >>>>     >sampling is still a viable solution.
>>>     >>>>     >
>>>     >>>>     >
>>>     >>>>     >Lastly, my rule of thumb for the filtering is Gaussian
>>>     3/3 for TV
>>>     >>>>     output and
>>>     >>>>     >Mitchell 4/4 for film output. Gaussian gives a blurrier
>>>     sample
>>>     >>>>     >interpolation, which may "conceal" minor imperfections.
>>>     >>>>     >
>>>     >>>>     >
>>>     >>>>     >
>>>     >>>>     >Cheers
>>>     >>>>     >Bernard
>>>     >>>>     >
>>>     >>>>     >
>>>     >>>>     >
>>>     >>>>     >
>>>     >>>>     >On 6/1/06, Helen Bucknall < helen(at)helenbucknall.com
>>>     <mailto:helen(at)helenbucknall.com>
>>>     >>>>     <mailto: helen(at)helenbucknall.com
>>>     <mailto:helen(at)helenbucknall.com>>> wrote:
>>>     >>>>     >
>>>     >>>>     >
>>>     >>>>     >>Mostly I just use min -1 and max 2. and box filtering
>>> 1 or
>>>     >>>>     sometimes I
>>>     >>>>     >>use triangle 2 for a softer effect.
>>>     >>>>     >>
>>>     >>>>     >>I output a string of 852 x 480 bmps to my clients and
>>> then
>>>     >>>> they mpeg
>>>     >>>>     >>it somehow and wack it up on big plasma screens. They
>>> never
>>>     >>>>     complain.
>>>     >>>>     >>When they want stills for print I just render a frame
>>>     out huge
>>>     >>>>     >>(3000+)but leave the antialiasing the same.
>>>     >>>>     >>If it's for telly I do HD res. If it's to be comped
>>>     with live
>>>     >>>>     action I
>>>     >>>>     >>ask what anti aliasing they want.
>>>     >>>>     >>
>>>     >>>>     >>
>>>     >>>>     >>
>>>     >>>>     >>Bernard Lebel wrote:
>>>     >>>>     >>
>>>     >>>>     >>
>>>     >>>>     >>
>>>     >>>>     >>>Samples 20 is a bit low. I generally put 64 as soon as
>>>     I set
>>>     >>>> up the
>>>     >>>>     >>>shadow map.
>>>     >>>>     >>>
>>>     >>>>     >>>What do you mean with "standard antialiasing"?
>>>     >>>>     >>>
>>>     >>>>     >>>
>>>     >>>>     >>>Bernard
>>>     >>>>     >>>
>>>     >>>>     >>>
>>>     >>>>     >>>
>>>     >>>>     >>>
>>>     >>>>     >>>On 5/31/06, Helen Bucknall < helen(at)helenbucknall.com
>>>     <mailto:helen(at)helenbucknall.com>
>>>     >>>>     <mailto: helen(at)helenbucknall.com
>>>     <mailto:helen(at)helenbucknall.com>>> wrote:
>>>     >>>>     >>>
>>>     >>>>     >>>
>>>     >>>>     >>>
>>>     >>>>     >>>>OK. I dragged the spotlight in closer to the scene
>>>     elements
>>>     >>>> and it
>>>     >>>>     >>>>seems to have stopped all the flickering! It wasn't
>>>     that far
>>>     >>>>     away,
>>>     >>>>     >>>>but now it is almost close enough to make my character
>>>     >>>> sweaty. I
>>>     >>>>     >>>>also scaled down my ground plane a tad.
>>>     >>>>     >>>>
>>>     >>>>     >>>>Just using 1024 res and sample 20. And bog standard
>>>     >>>> antialiasing.
>>>     >>>>     >>>>
>>>     >>>>     >>>>So very BIG THANKS to all who replied!
>>>     >>>>     >>>>
>>>     >>>>     >>>>Joe Laffey wrote:
>>>     >>>>     >>>>
>>>     >>>>     >>>>
>>>     >>>>     >>>>
>>>     >>>>     >>>>>On Wed, 31 May 2006, Helen Bucknall wrote:
>>>     >>>>     >>>>>
>>>     >>>>     >>>>>
>>>     >>>>     >>>>>
>>>     >>>>     >>>>>>Is there a way to get shadow map shadows to stop
>>>     >>>> flickering even
>>>     >>>>     >>>>>>when nothing is moving?
>>>     >>>>     >>>>>>
>>>     >>>>     >>>>>>This is what I have tried to no avail.
>>>     >>>>     >>>>>>
>>>     >>>>     >>>>>>On the spot light (cone angle less than 90):
>>>     >>>>     >>>>>>Resolution up ( and down).
>>>     >>>>     >>>>>>Softness up and down.
>>>     >>>>     >>>>>>Samples up.
>>>     >>>>     >>>>>>
>>>     >>>>     >>>>>>None of these make any diference to the flickering.
>>>     >>>>     >>>>>>
>>>     >>>>     >>>>>>On the render panel:
>>>     >>>>     >>>>>>I've played around with the antialiasing and jitter.
>>>     >>>>     >>>>>>On the shadows page I've clicked off rebuild.
>>>     >>>>     >>>>>>
>>>     >>>>     >>>>>>
>>>     >>>>     >>>>>
>>>     >>>>     >>>>>If you view your scene from the light is there an
>>>     >>>> extremely large
>>>     >>>>     >>>>>depth to the objects? In other words, are the closest
>>>     >>>> object to
>>>     >>>>     >>>>>the light and the farthest object from the light
>>>     extremely
>>>     >>>> far
>>>     >>>>     >>>>>apart? This can lead to ZDepth (used for shadowmaps)
>>>     >>>> presision
>>>     >>>>     >>>>>issues in some renderers.
>>>     >>>>     >>>>>
>>>     >>>>     >>>>>This is often the case if you have a huge ground
>>> plane.
>>>     >>>>     >>>>>
>>>     >>>>     >>>>>You may be able to exclude some objects, to help, or
>>>     make
>>>     >>>> smaller
>>>     >>>>     >>>>>versions that catch shadows.
>>>     >>>>     >>>>>
>>>     >>>>     >>>>>--
>>>     >>>>     >>>>>Joe Laffey                |       Visual Effects for
>>>     Film and
>>>     >>>>     Video
>>>     >>>>     >>>>>LAFFEY Computer Imaging   |
>>>     >>>> -------------------------------------
>>>     >>>>     >>>>>St. Louis, MO             |       Show Reel
>>>     >>>>     http://LAFFEY.tv/?e01274
>>>     >>>>     >>>>>USA
>>>     >>>>     >>>>>
>>

---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi