No Anamorphic Rendering ... Again!!!

Generic forum to discuss Octane Render, post ideas and suggest improvements.
Forum rules
Please add your OS and Hardware Configuration in your signature, it makes it easier for us to help you analyze problems. Example: Win 7 64 | Geforce GTX680 | i7 3770 | 16GB
riggles
Licensed Customer
Posts: 493
Joined: Mon Feb 17, 2014 3:34 pm
Location: CT, USA

pumeco wrote:The attached file isn't the best of examples, but you can see that Anamorphic bokeh look much more real-world lens like and cinematic than the standard bokeh. The problem with the technique used here is that the scene objects themselves had to be stretched, and that only works in a head-on view, that's why the film plane itself needs to be able to change it's aspect so that what is captured to it will get deformed in postwork when it is squeezed back, that's what happens in the real world, that's how it works. Basically, a cinema projector has an Anamorphic lens attached to it which is used to correct the Anamorphic lens that was attached to the camera they shot the film on.
My only thought is on this highlighted part. While this applies to working with film and projectors, that's not really how post-work is being done—at least not in my experience. Since all HD video output is square pixel, digital footage captured with anamorphic lenses is converted to square pixel during ingest. So edits and post work is done in square pixels. So the output of Octane shouldn't have a 2:1 pixel aspect ratio and have to get stretched back in comp. That would send us back to the SD/DV days, and we don't want to go there. :)
User avatar
pumeco
Licensed Customer
Posts: 61
Joined: Thu Aug 19, 2010 10:59 am

.

But that's the way it has to be done unless the wizards at Otoy can make it work passively somehow, so that we don't need to squeeze back in postwork.

Actually, that should be quite easily done, nothing hard about changing the aspect of something. Unfortunatley, I don't know the terms the Otoy developers would use, so the best thing I can say to Abstrax is that basically, take what ever is being rendered and render it as normal, but ensure that the last step (the virtual film plane), can have it's aspect ratio changed, because that way, whatever happens in the virtual lens behaves as normal but it is captured to an adjustable plane.

If they can then take that plane and recorrect it automatically, that would be fantastic, but not essential as it can be done in post.

.
Windows 7 64-bit | 2X GeForce GTX 460 | nForce 980a Hybrid-SLI | AMD Phenom II X4 3.40GHz | 16GB
User avatar
pumeco
Licensed Customer
Posts: 61
Joined: Thu Aug 19, 2010 10:59 am

.


BTW, Abstrax, this is done using a bokeh trick, but this is the effect you get with Anamorphic bokeh, watch how it behaves when defocusing.

Check this video out:
https://www.youtube.com/embed/66rrPfi1CG8


.
Windows 7 64-bit | 2X GeForce GTX 460 | nForce 980a Hybrid-SLI | AMD Phenom II X4 3.40GHz | 16GB
riggles
Licensed Customer
Posts: 493
Joined: Mon Feb 17, 2014 3:34 pm
Location: CT, USA

pumeco wrote:But that's the way it has to be done unless the wizards at Otoy can make it work passively somehow, so that we don't need to squeeze back in postwork.
Both V-Ray and Maxwell have been doing anamorphic bokeh without changing the pixel aspect ratio of the output image.
User avatar
pumeco
Licensed Customer
Posts: 61
Joined: Thu Aug 19, 2010 10:59 am

.

I think you just crossposted with me, check out that video I just posted, pretty cool example.
But yup, Anamorphic is needed, those other renderers have it like you said.

All we need now, is to get it in Octane ... muahahaha ... beautiful Anamorphic :twisted:

.
Windows 7 64-bit | 2X GeForce GTX 460 | nForce 980a Hybrid-SLI | AMD Phenom II X4 3.40GHz | 16GB
User avatar
abstrax
OctaneRender Team
Posts: 5508
Joined: Tue May 18, 2010 11:01 am
Location: Auckland, New Zealand

Ok, how about this:

I add two new inputs to the thinlens camera: One to control the pixel aspect ratio (to allow you to render non-square pixels like NTSC/PAL) and one to control the aspect of the DOF disc . This how it look like in action:

Pixel aspect 1:1 / DOF aspect 1:1:
pixel_1_aperture_1.png
Pixel aspect 2:1 / DOF aspect 1:1:
pixel_2_aperture_1.png
Pixel aspect 1:1 / DOF aspect 2:1:
pixel_1_aperture_2.png
Pixel aspect 2:1 / DOF aspect 2:1:
pixel_2_aperture_2.png
Pixel aspect 2:1 / DOF aspect 1:2:
pixel_2_aperture_0_5.png
Do you want to be able to set those ratios as fraction (i.e. two float values) or as a value (i.e. one float value)?
In theory there is no difference between theory and practice. In practice there is. - Yogi Berra
riggles
Licensed Customer
Posts: 493
Joined: Mon Feb 17, 2014 3:34 pm
Location: CT, USA

abstrax wrote:Pixel aspect 1:1 / DOF aspect 2:1:
pixel_1_aperture_2.png
Do you want to be able to set those ratios as fraction (i.e. two float values) or as a value (i.e. one float value)?
That's getting there, and much quicker than I thought :D

Slight difference, though, with anamorphic in that the difference between a 1:1 and 2:1 is that the DOF disc in the 2:1 image is shown here as half the height of the 1:1 DOF disc, instead of twice as high with an anamorphic lens.

I would think that since both pixel and DOF aspect ratios are both n/1 by nature, a single value would be sufficient for each. So bokeh aspect ration would be "2".
User avatar
abstrax
OctaneRender Team
Posts: 5508
Joined: Tue May 18, 2010 11:01 am
Location: Auckland, New Zealand

riggles wrote: Slight difference, though, with anamorphic in that the difference between a 1:1 and 2:1 is that the DOF disc in the 2:1 image is shown here as half the height of the 1:1 DOF disc, instead of twice as high with an anamorphic lens.
So just invert the horizontal/vertical stretching behaviour - like this? (pixel aspect = 1, aperture aspect = 2)
pixel_1_aperture_2_inverted.png
I would think that since both pixel and DOF aspect ratios are both n/1 by nature, a single value would be sufficient for each. So bokeh aspect ration would be "2".
Ok.
In theory there is no difference between theory and practice. In practice there is. - Yogi Berra
riggles
Licensed Customer
Posts: 493
Joined: Mon Feb 17, 2014 3:34 pm
Location: CT, USA

abstrax wrote:So just invert the horizontal/vertical stretching behaviour?
Yes.

There's a lot more physics going on in a real anamorphic lens, like it's unmistakable capturing of lens flares, and also it's change in effective focal length. But this is a very nice start in the right direction. Thanks for taking the time out to give this a whirl~
User avatar
abstrax
OctaneRender Team
Posts: 5508
Joined: Tue May 18, 2010 11:01 am
Location: Auckland, New Zealand

riggles wrote:
abstrax wrote:So just invert the horizontal/vertical stretching behaviour?
Yes.

There's a lot more physics going on in a real anamorphic lens, like it's unmistakable capturing of lens flares, and also it's change in effective focal length. But this is a very nice start in the right direction. Thanks for taking the time out to give this a whirl~
Ok, done (will be in 2.22). Anything more fancy will have to wait unfortunately, but I created a ticket so it doesn't get lost.
In theory there is no difference between theory and practice. In practice there is. - Yogi Berra
Post Reply

Return to “General Discussion”