Where is the secret lab, you say? It's RIGHT HERE. cheers, dr geep ;=]

Now that's the kind of experiment we need here. I strongly suspect that Poser's "camera" has its own set of formulae, which bear a only a vague resemblance to the formulae (or tables) for actual film cameras.

Yup, I 'spect you've hit the ol' nail right on the ol' snozolla. ;=]

Attached Link: http://homepage.mac.com/leemoon/DOF-HelperPoser5.pdf

Here's the PDF address:
http://homepage.mac.com/leemoon/DOF-HelperPoser5.pdf

Hope this is of benefit. :)
Lee

.

Wow! I think I died and have gone to Poser heaven. This is some great stuff folks. Thanks for the pdf LeeMoon, and Geep, its good to know the DR. still makes house calls. Now about that electronic surveilence equipment, what you want to do - voice fades to a whisper....put on some Pink Floyd, the Wall is a good choice, and point the speakers that way. Better than white noise any day. :O)

Hmmm ............... vedddddy interesting ! ;=]

Oh, and about that hou$e call ........ you'll get my bill. ;=]

That's good, and you will get mine for TSCM (er in layman's terms technical surveilence countermeasures, ie. debugging what bugs yah) :O) Strangely enough they should cancel each other out - what is commonly known in the TSCM field as nullification. I'd tell yah more but then...well you know the rest. ;O) On a more serious note, looking forward to seeing what else you can come up with on DOF. So far I am spinning my wheels and not making much headway on my tests. But that's why I am only Eeegor and you are the Doc.

Would you be willing to help me with an experiment? My thought is this: We can set up a poser cube at the origin, which would be the focus object, then set up additional cubes every 1 poser unit aligned to the axes of the ground plane (x and z). Then, by modifying the f-stop in prearranged steps, we can get a rough estimate of the "sharpness zone" in poser units. If we arranged that data into a tabular format, it shouldn't be too hard to use a spreadsheet to graph the information and come up with a "close enough" mathematical formula to describe the curve. In order to check our results, we could try the experiment several ways. My suggestion is that we would: a. Get the data for 10 separate f-stop values at 35 mm with a set distance frm the focal point. b. Get the data for the same f-stops at 70 mm with the same distance from the focal point. c. 10 renders, same f-stop values, 35 mm, double the distance from focus. d. 10 renders, same f-stop values, 70 mm, 2x distance from focus. That would require 40 renders in all, which could give us a reasonable data set to generate some real relational data. Can anyone think of a reason this wouldn't work? I can probably set the experiment up as a pz3, and host it on my server, so that anyone with a decent connection speed could contribute or try variations I haven't thought of.

Notes: the reason I suggest a cube is that we can get a good guess at the "sharpness falloff" based on the distortion of the visible edges, which we could reasonably agree to be the standard for measurement. ( If the cube on 0, 8 is fuzzy around the edges, then the sharpness measurement would be expressed as 8 pu's from focus to edge )

I think it a good idea. I was also going to suggest someone set up a pz3 and everyone use it to have a controlled test. Since you also thought of that all I can do is applaud you for your efforts and volunteer to help do the test renders. I think if we can furnish ockham with the right data he can make a script that will simplify using DOF in P5.

Oh, and I would make the instructions as to the how to steps to perform the test as "dummy free" as possible since I tend to confuse easy until things sink into my thick noggin. :O)

Setting up the pz3 now. :) I'll post when it's available.

Well, I have the basic PZ3 set-up, but there is a problem. (Actually a few)

First of all, the script is insistng on calling a figure (or possibly a prop that is attached to a figure). So, it won't set my Focal Distance against the ball prop.

I might be able to get around that by loading the DNA_Skydome, but that brings me too my next problem:

1. If I add a figure to the scene I can't redistribute the pz3. (I can use HER, but I won't without sixus1's permission - even if it is a free figure ... Same with the DNA Skydome)

2. If I add a figure, the pz3 will bloat pretty badly, making it a much larger download. That's not a problem on my end. I have more bandwidth and storage than I know what to do with, but it will make things a lot more difficult to download.

If I don't load a figure, then the focal distance and f-stop defaults are too messed up for me to get a clue as to what are reasonable settings to focus on the box. :(

I've not gotten a reasonable focus on the box after several tries (actually, several millions of tries)

stewer, if I'm reading your code correctly, the value I'm trying to establish (focal distance) should be determinable by calculating the resultant vector from the vectors formed from the difference in position between my camera and my focus on the x, y, and z axes?

If that's the case, then there has to be some offset value between the camera and the prop, or some variable I'm not seeing, because I don't think I'm screwing up the math. :(

Hm....digging deeply in my first Poser experiences, it may actually be that there is some extra offset to the main camera that my script is ignoring. Will need to do some research. As for having a figure - just load the ball prop, enter the setup room, leave the setup room. There's your redistributable figure :)

Good call, stewer. Thanks! looks around to make sure no one's going to get upset for us redistributing a primitive =D (It's really cool to know I was right about the math function you were describing, btw. Think I should write a note to my Physics professor to let him know that the only use I've had for his semester of hard work was a failed attempt to calculate the distance between a virtual camera and an imaginary ball? 8D )

Ok. I got the script to work, but the focal length it returns doesn't appear to be right. (It returns a value of 24.802 for focal distance, which, after 8 or 9 test renders at varying f-stops seems to put my focus well beyond the figure. :( I'll zip up the pz3 real quick so that you can check it if you want (or so that others can download it and experiment).

Scratch that... It's good now. give me a few minutes before I post the pz3. (the focal length is right, it's me that's wrong in this case) ;)

Ok. Here's the link to the .pz3:

http://www.studioverite.com/forums/freebies/focalexperiment.zip

The boxes are spaced .2 pnu apart. The white box in the center is the focus, with teach box alternating color.

Ockham and stewer's script returned a focal distance of 24.802 which seems right.

I set the render size up at 1200x1200 so that there'd be enough detail to see a noticeable change.

I figured that starting with an f-stop of 10.0 we can work our way down to 1.0 in whole steps. Try to make a best judgement on where the distortion starts and ends on both the horizontal boxes and the vertical boxes. The distortion should increase as the f-stop value decreases.

I'm going to warn everyone in advance that rendering those 21 boxes under full production mode (which is set up by default on the .pz3) may require some firepower. It's not too bad at an f-stop 10, but at 1 my machine was crawling.

The pz3 is only a 10k compressed, so everyone should be able to get it pretty easily, even on dial-up. :)

Enjoy! Note: I'm not sure if this experiment is valid or not. In order to get a mathematical idea of what's happening we might need to look at the amplitude of the effect instead of the area of the effect, which will be much harder to come up with empyrical values for. But I'm not going to write it off completely til I've run it a few different ways.

And btw, even if it doesn't work out, we can just give the ball linear control over f-stop and still have more information on how to use it effectively. =D

BOOKMARK

Is there any way to automate the script to recalculate and alter the DOF settings between frames on an animation? I'm not sure how deep the Python rabbit hole goes. I assume that Poser ignores scripts while rendering an animation sequence (?), but animated DOF would be suh-weet! So I figured I'd ask, just in case. =D

...gotta go to work...just stumbled on this...must come back...

looks around to make sure no one's going to get upset for us redistributing a primitive Nothing is being redistributed. Look inside the pz3, the sphere geometry is only referenced.

Great stuff.

About animating this - that'd be tough. By default, the depth of field parameter is not animateable. You could maybe do it with a python script that: 1. writes a partial pose file containing render settings with focal length (see ockham's script above) 2. renders the current frame, saves it 3. advances one frame 4. repeat until animation is finished

Got the test file. Must do a little work this morning. Will play, er experiment, later today. Thanks for setting thing up Softriver. Hope my little ol puter is up to the challenge. :O) Stewer thanks a bunch for your imput also. P.S. If I am sounding a bit MommaHennish it's cause I started the thread, and I feel "thanks you" are in order for everyone taking part. To me, this is the forum at its best. Not just moaning and groaning but finding ways to make Poser better - yeah just like in them good old days :O)

Yeah !!!

But, .............. just think ...

... at some time in the future ...

... These days will be ...

............................................... "them 'good old days.' "

Think about it. ;=]

cheers,
dr geep
;=]

( It's ok, shadownet, ... I miss 'em, too.)

Message edited on: 11/22/2004 09:47 ..... to remove all TOZ violations ... ;=]

Message edited on: 11/22/2004 09:48 ... to add the "to remove all TOZ violations" statement.

Message edited on: 11/22/2004 09:50 ... to add the "to add the "to remove all ... ... Oh ... nevermind. ;=[

Message edited on: 11/22/2004 09:53

The above post was brought to you by: Editor's Anonymous ;=]

he he, that Anonymous guy sure gets around

He gots lots 'n lotsa brothers. ;=]

Bookmark It's very interesting...

Yes, animating this would be coolest for sure - like the camera guy turning the focus ring slowly and switching focus from BG to FG, film style. Yummy.... :] Fish

Yes, it would be cool if the fstop setting can be animated. It would also make this test faster/easier as we could do a batch render frame 1-10 with fstop 1-10, etc. Remember Necessity is not the true Mother of Invention - laziness is. :O)

... and NSG is the father of "laziness." As a matter of fact, he invented it. ;=]

Ask NSG if he happens to be his father's son's mother's offspring on his daddy's side?

Don't hold your breath, ... ... it will probably take him at least, um, er, oh, about, um, er, ....... 5 years just to diagram that question ... so he can figure out what you "axed" him. BTW - He's "stoopid," also. But, you can learn more about that in Tutsin Comix, Issue #2. ;=]

Besides, NSG's father died tragically when he drowned in a small pond about 2 years before NSG was born so, he never met his father. His mother disavowed any knowledge of ... She used to watch alot of "Mission Implausable" on the TV. I think there might be a connection there. ;=]

DRAT! I always thought that was "Mission Implosable" (after watching those episodes where the plot collapsed in on itself). Thx for clearing that up for me..:-)) Cheers, Diolma

If anyone who has downloaded the test has managed to get through a series of the renders, please e-mail your results to me as a text file at softriver@studioverite.com)

So far, I've gotten through all four stages of the test with mixed results (I'll post them after I hear back from a few people).

Ockham, I was thinking about how to make this idea more intuitive. Instead of the target figure's scale, would it be possible to use (f-stop value = x rotation(in degrees)/10)? That would give us a value of 0 to 36.0 for f-stop (disregarding Poser's 6 sig digits) and we'd be able to "fiddle" the value with a linear dial twist.

I've come up with a few problems involved with the "scaled f-stop" idea. Mainly, getting a high f-stop value might require creating scene occlusion. Rotating the sphere in x, however seems more natural. stewer: Does Poser actually use -180 to 180 degree references? If so, linking to rotational value would mean including a line that flopped negative rot values to a positive integer, right? If that's necessary, I can post C++ code to do so, and you Python gods can just convert it if you like. =D

Just logged on and stopped by this thread to see how things are going. I can't believe how much you've all done. Pretty cool to see everyone with their sleeves rolled up. Glad I bookmarked ...

Hi softriver, taking me a while to render. I have the first 10 done. Starting on the next 10 soon. Also will be gone tomorrow so not sure I will get this wrapped up until Wed. Will try though. :O) Geep - tell NSG that's okay, cause he's stoopid and he knows he's stoopid. That puts him one up on a lot of other folks. :O) P.S. On the subject of his partenage - so that's what you've been doing in your secret laboratory. Let me guess, EEgor mixed up the bottles with the brains again. :O)

shadownet, You're close ................... REAL close. ;=]

I'm in the weeds here. Hey, what about using a ruler object running along the z-axis setting near the sphere or focus object. Wouldn't that blur outside of the focal area? One of the pages in the above link used a tape measure looking thing ... And what's the deal with Poser Units and English Feet and/or inches. I did a search and found a thread featuring our good Dr. Is one Poser Unit 8'?!!! I'll be here in the weeds ... Rance

he he Softriver - what camera settings are you using for the other 3 tests? I realize when I change the main camera to 70mm that I will need to adjust the DollyZ to see all the boxes. So will this through things off if we all do it differntly? If we wanting a controlled test, should we not all be using the same settings? Or does it matter? Just checking. Will not pretend to really understand what I am doing. Just rendering. LOL

Xpost - the he he was for Geep :O)

re: "And what's the deal with Poser Units and English Feet and/or inches. I did a search and found a thread featuring our good Dr. Is one Poser Unit 8'?!!!"

Uh oh, here we go ................. again.

I'm not gonna touch that one with a 10 Pnu pole.

Anyone ... anyone?

cheers,
dr geep
;=]

Sorry, Good Dr., I read further and further, and further. Shadownet, I found a post dated 2002, with you asking for someone to write a python script for calculate focal distance. http://www.renderosity.com/messages.ez?Form.ShowMessage=876578 I DO NOT WISH TO STIR UP THE POSER UNIT TO SCALE ISSUE here. I love and respect our good Dr. and have read a number of his tuts. Please forgive me for the earlier question. Mayhapp I should have stayed in the weeds! I usually don't post much here. I am more of a lurker as I lack the great skills of my betters. Thanks to all working on this project, Rance

Shadownet: For the 70mm tests (1st set) I kept the dolly - z as is. You can't see all of the boxes during those renders, unfortunately, but you'll still be able to see the results for the most part. The other option (which I also used) is to increase the aspect ratio of your render window. I went with 1200 px wide by 1600 px tall. When you double the dolly z, all boxes will be in clear view for both rendersets. Rance- I don't have a ruler object, and Poser's grids don't render for me. Maybe I'm missing something. Also, I want separate objects with no real textures applied to make the renders run as quickly as possible. You should realize that with the DOF settings on a very large render, my box is crawling already with 126 polys, three lights, and no textures (and it's a pretty 'notch pile of silicon). ;) Cheers, soft