I had completely forgotten about that. That must be back when I first got Poser 5, all happy and excited. Then I got really frustrated with it crashing on me and shelved it for about two years. Things have improved since then and I am now happy and excited again to be learning P5. :O)

Softriver, okay I just wanted to be on the same page is all since I was pretty sure moving the camera would change the render results from one person to another if we all did it differently. I will render 70mm with the dolly z as is. I do not understand the other two test. Please explain those. No point my doing the test wrong just cause I'm to thick to figure it out, and to dumb to ask - so I am asking. :O)

re: "I DO NOT WISH TO STIR UP THE POSER UNIT TO SCALE ISSUE here."

Nor do I, my good friend, nor do I ... but, fear not, for the good Doctor knows how to let a sleeping dog lie. ;=]

I was just having a bit of fun with you, ......... sorry 'bout that.

---

My dearest Rance01,

I would forgive you but ... there is no need. ;=]

You raised a perfectly valid point ........ and my simple response is:

If you use "Poser native units" (Pnu's) ...
which are the same as "Wavefront Object (.obj) units
... which are, also, the same between and betwixt Poser4 and Poser5 ... then there is neither a question of calibration nor conversion to anything else.

[caution - rant condition alpha is ON]

I use Dr Geep's Scale (DGS) because ... it works ... it is as simple as that, .............. period.... ;=]

Others may choose to use it or not ...
... at their own peril. (chuckle, chuckle)

[note - rant condition alpha is OFF]

cheers,
dr geep
;=]

I usually shoot at the tightest possible f-stop... but my usual subject is alpine wildflowers (10 violets to cover the face of a dime) and so maximum depth of field is more essential than light. It's just a longer exposure. I'll brace and hold, although I'm not as steady as I was a few decades back. Since fstop controls both the amount of light (exposure time) and depth of field (area within focus), it gets really complicated really fast. Film is sensitive to exposure times and will change color with longer exposures (something which won't affect us in this virtual world). However, there are some neat things you can do with real film (backlighted silhouettes and waterfalls flowing like milk and trailing stars) which are quite tricky to fake. If we can get Poser 5 to behave more like a real camera, I'll be tickled pink. So, thank you all! Carolly

Here's an approach to measuring or at least getting a feel for the DOF using the MeasureIt script (available in FreeStuff). I built a checkerboard grid in Photoshop (attached) and applied it to the ground plane. I used the script to measure the distance from the camera to the origin of the ground plane then used that as the value for the focal distance in the Render Options window. Setting this up is described in the next message.

Setup:

The Main camera's orbit and dolly values where all set to 0 and the script started. If you haven't used the script, it works by measuring the distance between the pointed ends of a couple of props built by the script. The props can be moved around and the distance between them is displayed in a TKInter panel. The Pointer2 prop was moved 1.1 PNU along Z, that's the distance from the front of the Main camera to the origin, and then parented to the camera. The Main camera was set to point at Pointer1 located at the origin of the ground plane. The attached image shows what this looks like using the Aux camera. Pointer1 is hidden in the image but is located at the base of the red '4'. The first test is in the next message.

The 50mm image:

This first render shows the result of rendering with the camera set at 50mm with an f-stop of 1.001. The odd f-stop number was chosen to give the most compressed DOF possible. To set this up, the main camera was x-orbited to -18 degrees. When the camera was ready, the distance from the front of the camera to the origin was displayed by the script and entered into the Render Options window. You can see a nice clean in-focus line that runs the length of the x-axis. Just to make certain that setting up the camera this way would hold up with an even more compressed DOF, I set the focal length to 1200mm and tried again. Results are in the next message.

The 1200 mm image:

The result here turned out better that I had expected. Again, a nice clean in-focus line running along the x-axis. You can also see how the smaller checkerboard has more squares that are going out of focus. By rendering larger images, you should be able to put a reasonable estimate on the actual DOF over a range of focal lengths.

I did find it a bit odd that FireFly rendered the DOF along the entire length of the x-axis. In a real photograph, the DOF should look circular with items near the center in-focus and items out on the edges not so. Guess that's a limitation of FireFly. I also found it interesting that the DOF is apparently the same distance in front and in back of the focal point (in this case the origin of the plane). Usually the DOF is greater behind the focal point. It may be interesting to try a short animation using frame by frame distance values by this method. If I work one up I'll post it.

::: Watching in awe ::: Coolest thread ever

Bookmark

Great info oh Divine Bushi, whose name must also be whisphered with awe. What you said about the DOF apparantly being the same distance in front and in back of the focal point is what I have been experiencing, and hoping to overcome. If this is a limitation of firefly than it would explain why so far no settings I have tried has allowed me to get the kind of gradient fade I had been hoping for. sigh.

Is that the best way that the depth of field can be rendered? it looks a bit low resolution.

@xantor - No, the renders where done with the pixel samples option set at the default value of 3. Higher settings greatly increase the render time but the out of focus areas look more realistic. I have a render cooking with the sampling set to the max (12) and I'll post it when it's finished.

Here's the render with the pixel samples option set to the maximum. There is still some grain but this was rendered directly to the main window. If the render was done to a new window at double the current size then reduced, the grain would be minimal.

bookmark

Could some one test this for me. So far the few minor test I have done have worked, but that is mainly on the WIP that launched this thread. I am in the process of doing a full blown render so my computer is tied up at the moment, but this seems to work.... What I did - and this use to drive my math teachers crazy, and I was okay until they asked me to SHOW my work. grrrr! I always got the right answer, but they marked half off for doing it the wrong way. grrr Okay, here it is in a nutshell Figure I want in focus is at the following x, y, z z=2.021, x & Y = 0 (this is the BODY locations, did not use the hip to place, but would be similar) Here is my Main Camera at 55mm x, y, z Dolly z= 9.617 y= 3.019 x= .246 Now cause I suck at math, I did the following and roughly added the x, y, z, camera values together and came up with the sum 13 rounded off From this I subtracted the figure Xtran 2(.021 dropped) and came up with 11 Next I open the render option and set the fstop to 16 - why, because dimly in the back of my mind I seem to recall that f16 was a good general setting for the kind of DOF I wanted. In the focal distance slot I insert - yeap - 11 I now render, and the render turns out right. I repeated this using a couple other items in me scene and seemed to get similar results. Now I am wondering, is this a Fluke? Or is it just that easy? [scratches head and hopes Poser Gods do not cast lightening bolts his way]

Oh, one thing I should clarify. The reason I subtracted the figure xtran is because it was a positive value like the camera dolly x, thus putting it closer. Had it been a negative value, thus farther away, I would have added, in order to get the approx distance from the camera to the figure. No, it probably does not make sense to anyone else but me. :O)

Book Mark*

Another way of expressing the formula for the distance between two points is:

the square root of ((x2-x1)**2 + (y2-y1)**2 + (z2-z1)**2)

So plugging in your values gives:

the sr of ((0.246 - 0.0)**2 + (3.019 - 0.0)**2 + (9.617 - 2.021) **2)

the sr of (0.061 + 9.114 + 57.699)

which gives 8.178, so I'm not at all sure why your value of 11 works. Did you try it with a small f-stop number to see if it still works with a smaller DOF?

Hi bushi, I did some limited tests using the box set up softriver posted at various fstop settings. It seemed to be working, but how well I am not sure. I picked different boxes at different locations and did my bad math to come up with a sum to plug in. Again, in my strange little world of warped thought, it makes sense and seems to work. Whether or not it will work for others or is anything more than just a fluke remains to be seen. As for the math you cited, I will try a value of 8.178 in my art render (which is where the original values I cited came from) and see what happens. Maybe I just lucked up and was in close enough focal range not to tell the difference. Shrug. Me and NSG have a lot in common at times. :O)

@shadownet Hey, if it works for you then it works fine. :-) The formula I used is the same as in the MeasureIt script. That doesn't say though that it's the ultimate answer here only that the formula gives the distance between two 3D points. Which points to pick is the hard part ...

I agree, if it works it works. My logic is that the main camera zeroed is at 0 x,y,z And when I load a figure its default is normally 0 x,y,z So if I move the camera or figure away or in toward, i.e -0 or +0 that is the difference in the distance. Since reality is often a product of our own mind, and since I have been accused at times of living in my own world - this logic makes perfect sense to me. LOL Even better, it seems to work. :O) Which, btw, 8.178 did not, and my figure ended up out of focus.

Hmmm ... re-reading the discription of your original setup I can see why 8.178 didn't work for you. That calculated distance was from the camera center to the point located at 0,0,2.021 which is on the ground plane. Your intended point of focus was up off the gp probably the figure's face. That would change what the second point for the formula would be. If you would and if you still have that original setup, would please check the camera dolly xyz locations and if there are any orbit xyz values. I'd like to try that with the MeasureIt script after this last render is finished. It's been going about 4 hours now but is almost done.

@bushi, (crawls prostrate warily back toward the altar thinking that was really spooky) Right after my last reply, 3 tornados touchdown around us, the power went out, so naturally I lost my nearly finished full blown render in process. Will have to start that one over. Did not loose the file though, so have the settings for you. Yes there were some orbit values. Main Camera 55mm DollyZ = 9.617 DollyY = 3.019 DollyX = .246 z,y,x & scale (all) 100% z&x orbit = 0 yOrbit = -41 Figure Body x,y,z Rot = 0 xTran = 2.021 y&ztran = 0

This is a low render version of the picture that sparked this thread, for anyone who might be interested. Wish it were better for such an auspicious occasion, but from such lowly beginnings still great things might come as far as end results on the DOF study.

This was rendered to the main window at fstop 16, focal distance 11, paint bucket at 5 (the fade is still a bit grainy) hoping a larger render with higher bucket setting will help.

@Shadownet- If you could do me the kindness of posting the position of the mirror, and the bearded cupid figure, I'd appreciate it. Your render has sparked me with a strange idea, and I was wondering if Poser has the internals to pull it off. (i.e. What if the reflected image is in the focus area, but the reflected object is not?) Maybe I'm making things more complex than they are, but it seems like there are a lot of factors on how the reflected image would display in ray-tracing versus real life. (i.e. Does Poser use the distance from the mirror to the camera to determine the focal length, or does it use the length of the ray (object to mirror to camera) to make that determination?) Looking at your image, I think Poser uses the length of the ray, but it's hard to see clearly. On the larger hi-res version, look in the background of the mirror, behind the short fat guy. Is the background blurry? @Bushi - I've been having trouble with the focal distance spreading out all along the x-axis as well. It's really confusing me. In real life, there should be a sphere around the camera lens with radius=focal length. But with Poser, it's just a flat plane running perpendicular to the vector described by the camera's facing. This is a serious limitation in certain types of images, but maybe it can be put to good use in another way?

Softriver, okay will get those setting for you shortly. The best hi-res version I could get so far was with a pixelsize of 5 at 800x600. Not very hi res, but my ol puter puters out. Hard to tell off it with any real certainty, but it does not look like any image blur in the mirror background. Even in this pic the DOF is not that good, but is the best I have gotten so far in all my tries. At least in this one, using my crazy method of calculation, I was able to keep the foreground and main figure fairly crisp, with the background fading. I think Dante may have had Poser in mind when he penned The Divine Comedy.:O( Oh, about your tests. I have the first 10 done, and 9 of the second test. Not been able to do the 70mm 1 fstop test render cause my puter gives up and P5 just reaches a point in the render and stops like it is done, only it isn't. I think it is time to upgrade. Not had time to try the 2x the focal value test, but have a feeling my puter is not going to be able to pull it off.

@softriver Mirror Y rot = -8 x rot = 0 z rot = -3 xtran = -0.542 ytran = 0 ztran = 1.402 (all scale - if needed - 100%) Cupid y rot = 44 x & z = 0 xtran = -1.308 ytran = 0 ztran = -3.105 Yscale = 70% all other scale 100% Oh, error in comment about about paint bucket size 5, er, that should be pixelsize 5. Duh. So, is any of this DOF beginning to make sense to you. Except for my bit of crazy calculations, how Poser does the DOF does not make much sense to me or seem to resemble what I remember of real world camera settings.

OK, this is an example of having the pixel sampling set at 12 and the image rendered to a new window at twice the size of the final. It was reduced in PS with a little croppping off the top. The render time on this ended up being around 5 hours with most of the time spent rendering the bottom quarter. With the max pixel sampling and reduction of any left-over grain by resizing, FireFly can give a reasonably simulation of DOF.

@shadownet - Thanks for the info on the camera settings. I'll try them out later today. Oh, and keep your head down! :)

That looks really good bushi! I wish I could render past pixelsize 5 800x600, but anything higher than that and P5 just stops in mid render like its done. Oh, btw, going back to the 8.178 at post 120. I tried that setting again, only this time with an fstop of 32. Worked. My calculation, however, at the same setting was a bit more blurry on the figure, unlike 16 and 8 fstop. There is definitely a bit of Poser weirdness going on here in how P5 does DOF, but I am encouraged by your results. :O)

Attached Link: http://www.shuttercity.com/DOF.cfm

Hey All, Travelled today (busiest day of the year they say). Rought flight too. Came across a link from, I think, RuntimeDNA, to this site http://www.shuttercity.com/DOF.cfm Kind of a neat do-hicky. Don't know if this helps or not. Best Wishes and Happy Thanksgiving, Rance

Hello Everyone, softriver, I've taken one of my older files and modified it a bit for this forum. It actually included a mirror so I thought I'd do a little test. Here's the stats: DoFTwo Main Camera Focal 55mm xTran 1.354 yTran 3.665 zTran 9.514 xOrbit -6 yOrbit -25 Mirror xTran 4.499 yTran 4.340 zTran -5.867 yRotate -1 xRotate -5 Don (Body) xTran 5.849 yTran 0 zTran -1.178 yRotate -158 Production quality, FireFly Use Dispacement Raytracing Smooth polygons Pixel samples 12 Depth of Field F-Stop 1.4 Focal distance 26.48 I used Stewer's script to calculate Focal distance. The default F-Stop of 2.8 didn't generate much effect so I chopped it in half. The image was rendered at 800x600 pels, .TIFF and resized to 600x450 and saved (20% compression) to .JPG btw, the chest figure/prop is DAZ I think.

bookmark