synthetic photography

[yonosoy]

13 hours over a pure 8K 16:9 format. Supersampling 3. Downsized to 4K with a 92% of compression.
MTL parameters : large mutation probability 0.72, maximum change 0.017
Canon 24 mm objective with a film of 20 milimeters width.

It seems like a mixture of paint, photography and cartoon. Is a perfect marriage between art and technology (my feelings complain against the diningroom as an artistic field, but well, all the intentions are there...).

MTL is image size dependent. Over a 8K format it gives a perfect photographic grain. In a range of 30-50 hours of processing it is possible to obtain a good 4K photography.

First attempt over this interior. Simple. Texture are free from internet, not the best at this moment. First attempt... A day of work with render time included. High contrast for the tones of the furniture that gives a dark image in general. Sun & Sky as lonely light source.

This exercise gives me the possibility to train for a possible professional collaboration to offer this workflow...

Canon 24 mm v01.rar

C4D scene, Houdini file.

(2.96 MiB) Downloaded 1220 times

[yonosoy]

8K, format 1:1, 24 mm film. A better focus. The spherical lost of focus is very interesting. There is some kind of optical vignetting, the diaphragm is 2º opened (some milimeters, is difficult to determine the correspondent F-stop).

process image, 45 min.
Left (processed), Right (EXR indigo´s output).

15 hours.

15 minutes, 4K, format 16:9. Indigo´s camera...

[yonosoy]

Look how the light changes between the use or not in the scene of glasses (sgg planilux).
Sadly there are problems with the primary ilumination. Specially with fabric material.

Here 5 minutes of the scene with indigo´s camera, there is not so brutal difference (maybe some more light in the second take above) and there is no problems with materials and primary ilumination...

[pixie]

Great work! As always!

[yonosoy]

Thanks pixie!
I´m not very happy with this actually... Can´t find scenes of my taste for test.

It may be difficult (I wait a lot of years for the connection of Indigo with Houdini), but I will start to model my own scenes to feed the thread. More than difficult it may be slow...

Not happy too with the actual situation of this kinf of excersice. Before some time working with minor problems, now I can not install glass panes, the inter-lenses vacuum don´t work anymore, and the cam gives some kind of brightness from the diapraghm in the center of the photographies... I´m modelling a new objective to test the behaviour. I must try to fallback to a precedent version of indigo to ensure that all is OK.

Will see, thanks again for the consideration!

[yonosoy]

Sigma 56 mm. Pure 4K, format 16:9. Without post-production. One hour each.

Warning: camera contained in more than one medium, forcing empty/vacuum medium.
This message appears when I put inter-lenses vacuum in the scene. The aspect of the image is this:

Without inter-lenses vacuum:

The mix:

The aperture:

[yonosoy]

This objective is more correct.
Inter-lenses vacuum installed with another procedure.
24 mm vertical film. Corrsponds with a 90 mm cinema´s camera, so let me investigate a little more.

9:30 hours, 2000 samples/pixel.

Sigma 56 mm.rar

Cinema file, Houdini file, Allembic file.

(1.06 MiB) Downloaded 1181 times

[yonosoy]

Now corresponds with a 72mm, perspective mostly corrected. But, the IOR of the flint glass is impossible (1.28).
Don´t know...

8K, downsized to 4 K, 800 spp, 12:30 hours.

[pixie]

Interiors should be hell to test this kind of lens... wouldn't it be wiser to use an external scene to tweak the lens and see how they work?

[yonosoy]

Maybe some nature?
I don´t know if atmosphere works now for lens systems (with a mesh as sun yes)...
With a little bit complex houdini´s vegetation and landscape and some ISL (so many times... Sergio always want to share the workflow..., maybe it is the time to do it).

Maybe I will try something...
But interiors is not exactly a hell actually.

Maybe it is important to explain in detail the model, assemble and use of lens systems. And to shown how it work!
Maybe it is a crazy fall in love and the differences that I experience with synthetic photography don´t deserve the effort...

[pixie]

I say that because interior scenes are notorious slow to render, GPU don't help as much as it could either since you cannot use portals, even with lights as windows is a bit slow from my experience, while exterior scenes are fast as hell.

Maybe it is a crazy fall in love and the differences that I experience with synthetic photography don´t deserve the effort...

That's good love!

[yonosoy]

Even with 13 million of instances for the carpet, interior is faster than a simple plane in exterior with sun & sky system (captured window, same position).

This is a 5 hours render of lens flares, with a LUT that camouflage the noise. In this case (only extra atmospheric sun), the speed is more than double the interior. I found this last render very noisy (maybe MTL values).

I´m agree with you, if love exists, is somewhat something like this...
It can let you as a fool...

The pattern is more interesting than what I obtained till now with the canon 24 mm.

[yonosoy]

MTL values: large mutation probability 0.2, max change 0.01

4K, downsized to 2K, 15 hours, 1400 spp.
Definitively, is better to launched the image in 8K and downsized to 4K. MTL is image size dependent...

Version 02 of the objective Sigma 56 mm. Now the perspective is correct, but the flint IOR is 1.24, and this seems to be not correct... Maybe there are another pair of IOR (of crown and flint) more correct. You must reconnect all the allembic objects.

sigma 56 mm v02.rar

Houdini file, allembic file, Cinema 4D file

(959.25 KiB) Downloaded 1170 times

[yonosoy]

I see now that the perspective is the same of the last post.
But in cinema 4D the consideration of the camera change from 100 mm to 59 mm. Don´t know why...

Maybe is because the format is vertical?
complex... For the protoestudios thread, the format is horizontal and now is correct, again, don´t know...

The format (vertical horizontal) is compensed with the film size, so is correct now.

[yonosoy]

In order to fine focus, you must move forward and backward the "enfoque" group.
If you do this, don´t forget to adapt through points the "vacio camara" and "cuerpo" components. They are selected, you only must move they. I think that the two groups of lenses inside "enfoque" don´t moves at the same time, don´t know.