[kami]

Hello everybody.

I always had the impression, that the Maxwell DOF isn't corresponding the DOF, my DSLR creates. Because with the normale scale of objects it was nearly impossible to get a really strong DOF. (I only got a decent DOF when scaling down the scene by 0.1).
So I tried to create a simliar testing scenario for my camera and maxwell render. These are the results...
The scenario: A bunch of rings, each with a distance of 15cm, starting at 30cm from the camera. At the end of the line: a black box.
Film size is 14.8x22.2mm, Focal length 45cm, Lens length: 50mm.

Both images with f/stop 1.4. On the left my DSLR, on the right Maxwell:


And both images with f/stop 2.8:


I think my assumption was right. The Maxwellian DOF isn't nearly as strong as the one created with a real camera. What is the reson for that?

Greets,
Kami

[jespi]

Have you checked if the global unit are correct?. In some 3d app, the internal unit can be different from the file units, like for example within 3dsmax. But maybe you are correct and your guess is correct.

[Mihai]

If you can upload the scene file somewhere it would help. You can upload it to divshare.com for example.

[kami]

of course i can upload the scene:
http://deiters.ch/maxwell/dof.mxs

I hope the scale is correct. I created it with the rhino-plugin (working unit was cm). I'm not that much of a studio user ...

to post the correct settings for my DSLR:
The pictures were taken with a Canon350D and a 50/1.4mm lens.

[Mihai]

The scale looks correct. I just loaded the scene and rendered and it looks a lot like the DOF from the 1.4 lens:



Did you try opening it in Studio and check? Was your render directly from the plugin and you uploaded the resulting mxs that the plugin created?

[Thomas An.]

I created it with the rhino-plugin (working unit was cm). I'm not that much of a studio user ...

Can you also upload the rhino file ?

[kami]

Hi. Yes, that looks a lot more like the result I expected
I rendered directly from rhino and uploaded the resulting mxs. I'll provide the rhino-scene as soon as i get home, this evening.

Update:
here is the Rhino file:
http://deiters.ch/maxwell/dof.3dm
I'm not sure which f-stop i chose last, so it's maybe not corresponding to the renderings posted above. but all other settings stayed the same.

[JDHill]

I'm not sure what I'm looking at here. I don't know how Mihai got the image he posted, that's not what I get using MXST/MXCL 1.6.1. First I opened up the dof.mxs you posted, to see the settings. Then I opened your dof.3dm and adjusted the camera to match (fStop was different), and reduced the resolution. This is how it rendered in MXCL, right out of Rhino:



Then I opened the .mxs in Studio and hit render:



Then I opened Studio's tmp.mxs and hit render:



Then I opened the .mxs you posted and hit render:



Finally, I opened the tmp.mxs from that and hit render:



I'm not saying this is any kind of answer...I am not yet sure what is going on. It looks like Studio is having some problem reading the file, but then it writes the _tmp.mxs, and reading that it's fine. Not to say anything about your original DOF-accuracy question - I don't think any of the above can be used as a good basis for making any judgement about that.

JD

[Mihai]

I opened and rendered in 1.6.1 as well. The skydome was much too bright for the camera settings so I just changed the ISO/SS in MXCL....

[Thomas An.]

Very interesting. I get the same stuff as JD.


About the DOF: Kami, what kind of camera are you using ?

[kami]

Hello. I didn't try to open the mxm's. All renders are from Rhino directly.

The camera i used was a Canon 350D.
Sensor size: 14.8x22.2mm
with a 50/1.4 lens.
focal lenght: 45cm, f-stop: 1.4
So this all matches exactly to the Rhino-Settings.

[kami]

any news on that subject?

[Thomas An.]

any news on that subject?

I think you might be on to something. The DOF used to be well tested. Maybe it got broke along the road ?

[blueplanetdesign]

Kami,

I followed the posts on the dof issue here a few days ago.

Some time ago I noticed the same inconsistencies and sort of let it fall by the wayside.
During some experimentation I did notice that the diaphram setting had a lot to do with the DOF.
Perhaps you ought to experiment with the camera diaphram settings and see if the
Maxwell rendered image comes closer to your digital camera output.

This pic uses a polygonal diaphram:


This pic uses a circular diaphram:


Both renders have an identical camera setting where the focus is on the nearest teapot. Lighting and other factors are also identical.
bpd

[tom]

During some experimentation I did notice that the diaphram setting had a lot to do with the DOF.

In fact, it wouldn't have such a big influence unless there's the following bug with 1.6.1
http://www.maxwellrender.com/forum/view ... hp?t=26925
(Btw, this is already fixed in the next version.)