Attached Link: http://www.renderosity.com/messages.ez?ForumID=12356&Form.ShowMessage=2573899

I'm posting again at top level because I bet some people missed it, so this will get it noticed.

There are nine more steps, so please try to wait for the next nine posts before asking questions, which I will be delighted to answer.

This tut will focus on infinite lights. Similar techniques will be needed for spots. IBL and point lights will not be covered here.

Tip: Check your Focal length. Too small and it distorts the figure. I like 150 mm.

Now delete all your lights.

You can use other orientations, but try this for now.

The light should be red,green,blue all at 1.0
Intensity 100%

Render.

Look!!! No nostril glow. But the shadows are fuzzy and indistinct. We'll fix that next.

So for now, right click on the camera selector in the preview window and choose the shadow cam. Don't worry about what number it is. I don't know why poser starts numbering from how many were in when you started poser. For me, there were three lights in the beginning, which I deleted. Anyways look through that camera.

Also, select the camera in the parameters window.

You should see something like my image. If your figure is not in the center of world space it may look different.

IMPORTANT: There is a wierd bug regarding shadow cams. After inserting a light, they do NOT show what they are actually looking at until you do at least one render. That's important!!! Whenever you add a light, before trying to work with the shadow cam, do a quick render. Even if you render nothing, it will "set" the camera properly.

For my picture of V3 in default pose, scale=12% works good for the head and neck.

Now as soon as you set the 12%, the camera will show a closeup of her feet. I don't know why, but the camera zoom always moves the camera to point at global coordinates 0,0,0. We need to adjust that.

For my figure, a yTran=510 works well. Sometimes you move xTran a little. Just get the portrait part of your figure completely in the image.

We want to make sure the area of interest is as big as possible, without leaving anything out. If you zoom too far and render, you will see the shadow mysteriously cut off at an arbitrary point. That means that your face camera can see something the shadow camera doesn't see. If that happens, scale up a bit and recenter the shadow cam.

This is what I get. Nice clean sharp shadows. No light in nostrils. Teeth get some light but have a nice shadow. Good shadow under the upper eyelid onto the eyes.

Add your light. Set it to infinite. Set min bias as before to .1.

Set the map size to 1000.

RGB 1.0,1.0,1.0
Intensity 100%

The orientation should be close to opposite the camera. Here I set yRotate to -160.

Remember: DO A QUICK RENDER FIRST. Otherwise you will be wasting your time.

Here I'm looking through my shadow cam lite 5 (actually the cam for light 2 - still with me??)

Adjust scale and ytran as needed. For me, the backlight worked at 13%, ytran=510.

This concludes the tutorial.

I DON'T WANT TO SEE NOSTRIL GLOW IN THE GALLERY AGAIN, OKAY!!!!?!?!?

Great stuff, thanks for posting. What kind of render time did you experience with the above shot?

Great tutorial, thank you. One question: what are the pros and cons of using your technique with depth-mapped shadows vs. ray-traced shadows? I've tried your test set-up with both and seem to get better results with eyelashes using ray tracing, although the shadows on the eye itself and the teeth seem better with the depth-mapped version.

1 minute

The neat thing about depth mapped shadows is that you can calculate them once and just reuse. So my workflow is much improved by them. Position my character, props, and lights. I turn on the "Reuse Shadow Maps" option in the render menu. Once created, the shadow maps get reused in subsequent renders so it saves a lot of time. I can then change materials, light strengths and colors, and camera position, without having to recalculate the shadows. If I'm using the raytraced shadows, I experience the extra calculation time associated with the shadows on every test render. Of course, you could turn them off, but if you are trying to get the lighting/mat/camera angle just right, it slows you down.

So given the somewhat bad results and the increased times, I choose depth-mapped 99% of the time.

Does the "Reuse Shadow Maps" option work when rendering an Animation? If the subject is moving then the shadow would change as well.

Correct - you can't reuse shadow maps for animations. I don't animate so I'm unclear on whether it even pays attention to that option. It shouldn't. By definition, animated figures will require a new shadow map each frame.

Thanks, that's useful.

when animating poser automatically re-renders the shadow map, even if you have "reuse" selected.

Thx, great tutorial, very helpful!

Just a minor point, which is obvious really, but if you alter the light position you will have to realign the shadow cam. For example, changing the light's xRotate to -30 instead of -20 throws the shadow cam right off, and the radioactive nostrils come back. This can be fixed by re-adjusting the shadow cam position.

for an animation, you'd have to animate the shadowcam. I find that with sufficient bias and map size, under depth map shadows, the nostril glow simply is not a factor; it is taken care of by the tight settings. However, I DO zoom in on the shadowcam for other reasons. ::::: Opera :::::

Thanks! This is very usefull information.

Thanks for the tute, bagginsbill.

Clear, concise and very informative minitutorial. Thanks!

Thanks for the tip!

Thank you for another great mini tutorial. Very useful.

Great ! Thank you !

bagginsbill, I was going well with this tutorial until step #6 of this thread, where you scale the shadow cam down to 12% and then move it up to the subject's head. When I tried this with the y-tran value suggested, the subject completely disappeared off the radar - of course she was still there, but the shadow cam had gone off in a different direction I guess. I've got two theories about this : (1) the positioning of the light associated with the cam; or (2) the units of measurement that I was using e.g. maybe I am not using Poser Units ??? Also, looked in the trusty Poser Reference Guide for some further enlightenment about the Shadow Min Bias, and the only description is as follows : "The Shadow Min Bias parameter specifies how far to shift samples towards the light source to prevent self-shadowing of objects." What on earth does that mean ? .... are you able to put this in more plain English ? thanks, Andrew

working with the shadowcams is difficult. There used to be, and sometimes still is, a dial "zoom" that appears. For the life of me, I can not figure out how to make that zoom dial appear, yet once in a while it does! This is a deep mystery. Meanwhile, if it does NOT appear, you have to manipulate the thing with only the x and y axes, plus "scale". It is no fun. About the bias, LOL, I'd love an actual explanation as well. All I know is, the lower the better. I use it, but I don't understand it. ::::: Opera :::::

Attached Link: http://physicalc-software.com/forum/viewtopic.php?t=20

Not sure if you were aware of this, but spots lights have a much lower less shadow map size requirement than infinites. For removing nostril glow, it's much easier to do if you use spots (you don't need to fiddle the shadow cams). Details above. Also, I totally agree with you - for production renders, depth mapped shadows are better than ray-traced.

Message edited on: 02/12/2006 16:23

Spotlight fan here echoing the sentiment. I would comment, however, that fiddling with the shadowcams has its own benefits: the closer in you zoom with the shadowcam, the smaller your shadow map can be and still attain similar quality results. This, in turn, leads to faster render times because the map is smaller. ::::: Opera :::::

Attached Link: http://www.renderosity.com/messages.ez?Form.ShowMessage=2581364

Attached Link: http://www.renderosity.com/mod/forumpro/showthread.php?thread_id=2581364

Also note that if your shadowcams are getting too out-of-hand, you have the following two options (could prove useful in animations):

• Lock the shadowcam. This is most useful if the light moves or points at different objects or areas instead of a single object or figure.

• Use point-at on the shadowcam. This is best for keeping it fixed on a single figure. You'll most likely want to point it at the head, chest or hip, depending on distance between light and figure.

In some situations, using both of these in combination may prove useful as well.

What renderer did you use P4 or Firefly?

Half of the time when I look through the Shadow Cam there is a blank scene. Does anyone know what is up or if there is a way to get around that? Some times I can get it back with a restart of Poser but not all of the time.

I discovered the nostrils have a selection in the Material room (P5), and I just use the 'bucket' to make them black..;) but I do everything the hard way.

GoM: whenever I see this problem, I click on the camera icon at the top of the preview window, and the shadowcam then usually pops back where it should.

I will try that. Thanks

Interesting stuff.  Thanks for posting it Baggins.

One question: what are the advantages of this approach relative to just using object-based AO on the nostrils body part material?

Thanks again... tguyus

Great tutorial...I'll have to try this out sometime!  Thanks! :)

I got tripped on the cam scaling too. As near as I can tell, Scale and Zoom are "the same thing." Nothing is really the same thing, of course, but they seem to do the same thing. It's possible that "existing" lights (the original 1, 2, 3, even if you changed them to spots) get shadow cams with Zoom and Pan, but new cams get the more primitive Scale and Trans. How do you tell an "original" cam? Its name is Shadow Lite # Cam rather than Shadow Cam Lite #. (Yes, I too wish I was making this up.)

And a potential cause of PanY whacking the camera off into some X: It wasn't PanY but PanX. Poser has a tiresome habit of presenting dials in whatever order it feels like, and when you are doing something like this, where you make "the same" adjustment to three or four things, it's a real PIA. My spot 1 shadowcam has Zoom, PanY, and PanX for controls, in that order. My spot 2 shadowcam has Scale, transX, and transY, in that order. After I moved cam 1 up to the face, I went to cam 2, dialed the second dial, and boink.

BTW, with Sydney G2, the first step -- fixing the shadow settings -- did nothing for her nostril glow. Not sure why. However, adjusting the shadow cams worked great. It fixed most of the nostril glow. I still see "light artifacts" that puzzle me, though. Some nostril glow bits (inconsistent and discontinuous) and light on her neck which should be in shadow. I'm suspicious that the IBL is causing them, so I'm running another render without it.

I have a theory question that's bothering me. I understood that the view from the shadow cam is a Light's Eye View of the scene, and I can see how Zoom and Pan would move the port around on the pane of the light. But they don't appear to do that, they appear to move the camera around on the X,Y global axes. Which is it? Is it Panning or Transforming. It looks like Transforming to me. The theoretical question: Does it not matter?

M
PS: I just did some testing and Yes, it does matter. The "Scale" shadow cams do indeed move according to the x,y coords rather than panning. To test this, move the camera slowly and watch whether what's visible in the center changes. With "Zoom" cameras the pane frame reveals new territory, but the plane stays the same. When you move a "Scale" camera across X, it behaves like an orbiting camera pointed at the scene -- which is wrong, isn't it?

An absolutely awesome tutorial Bagginsbill!

My question is, when would you have to recalculate a shadow map (ie; can't reuse map from previous render) other than animating? From what I'm understanding is that if I alter the strength, I don't have to recalculate. If I move the position of the camera, no recalc. If I move the light, recalc.
I'm just trying to make sure my head is wrapped around this properly.

Thanks again for your tutorial and time that you've spent on this.
Regards

This is Syd G2 with one active light. IBL off, Fill off. As the shadows attest, the light is off to her right, a bit above her head. The light's shadow cam has been adjusted and Lo, no nostril glow. However, the left side of her nose is being lit by... what? and the area of her neck in shadow has a stripe of light across it from... where? Apparently this is not the same problem as nostril glow. Does anyone have any idea what it is?

M

The light artifacts are mesh-dependent? I just put V3 into the scene, turned on all three lights, and rendered. It looks just fine. Same with Steph. And with Jessi, Ok too. So it's got something to do with the Sydney G2 figure. Arggghhh!

M

Apparently in P7 they change "Scale" to Zoom and XTran, YTran became PanX, PanY. Whatever they're called, I believe those dials work the same as P6. Are you using the parameter dials, or are you spinning the camera controls on top of the pose preview window? They work differently. The camera controls on the pose preview don't work right for shadow cams - they seem disoriented to the actual camera. If you stop and think about it, that makes sense as the camera is not actually free to move or rotate. That would involve moving or rotating the light, which if you try it, you'll see the camera moves with it.

What the parameter dials do is translate the origin and scale of the shadow map itself, in other words imagine a plane suspended in front of the shadow cam with perpendicular alignment.  Now imagine a sheet of paper in this plane. You could change the scale of the paper, or slide it around the plane. This "sheet of paper" is your shadow map! You're not actually moving the camera at all with the PanX and PanY - you're moving the paper.

And regarding the shadow problem - I believe P7 is broken. This is one of the reasons I won't use it. At RDNA we've had discussions of several other bizarre new shadow problems that P6 just does not have. Perhaps you, like me, should stick to P6.

There is evidence that P7 no longer obeys the Min Bias at low settings - no matter how low you go (.0001) there is still breakaway - as if the shadow is offset a tiny bit from where it should be.

Both Pan dials (original lights) and trans dials cause terrible fisheye distortion at 10-20% scale, which suggests orbiting to me. The fisheye effect makes it simply impossible to fit the cam to the area included in the shadow map.

And IBL w/AO screws up the shadows beyond fixing, as my Miki pic shows, even if the "Light" from the IBL zones is black. I've pretty much sworn off IBL; it looks like abandoning P7 is more in order.

Mick

Adorana posted mysterious bad shadows with IBL  at RDNA - she emailed EF and they acknowledged the problem. Don't know if the SR's address it.

Fish eye distortion is due to a low focal length on the camera. This is appropriate for spotlight shadow cams, as they actually are pretty darn close to the figure and have to be made wide angle to include the full spread of the light. If you narrow the width of the spotlight cone, the camera goes to a longer focal length. and there  is less fisheye. But don't worry about fisheye - you're not rendering the scene from that point of view - you're rendering the shadows and they are indeed fanning out in a cone.

Duh. I didn't think of the connection between the light cone and the camera cone.

I am wondering if the light artifacts (Monday 9:23 post) in the eF figures could be caused when by alignment of the normals. That scallop of light on Syd's chin really puzzles me.

The SR 1, in my experience so far, doesn't address any of the show-stoppers in P7 (like the knocking aside of a light when you click its globe to select it). Many of them were ignored in the service releases of P6, and some are left over from bug reports at the beginning of P5's life cycle. So I won't hold my breathe waiting for a fix.

M

 This didn't work for me in Poser 7.

In Poser 7, the shadow cameras do not have a scale parameter dial, they have a zoom parameter dial instead.

It would be nice to see a follow-up of this topic for Poser 7.