[Bf-vfx] Tomato camera sensor changes

[Ejner Fergo]

Hi Sergey!

> Looks like you misunderstand me.
>
> I'm not against adding vertical FOV. I just don't want to rough such
> changes. This thing should be introduced in nice way to fit:

Agree, but was the sensor patch really that bad?

> - Default Blender use-case (when it's used for 3d modelling and so). In this
> case i doubt user should care about sensor size, vertical FOV and so.

Exactly, no need to touch the camera settings, even if it is the most
advanced in the world.

> - Match-moving use-case (when we have to gather information about real
> camera and make match between modeled scene and footage). In this case
> information collected about real camera should correlate to what's supported
> with renderer. My project isn't related on improving renderer and so so
> currently i'm trying to stick to current design. Footage resolution + focal
> length + sensor width + pixel aspect was enough to make match between real
> camera and blender camera.

Yes and for match-moving the settings you included are fine (no sensor
height). This is inside the Movie Clip Editor, not the general camera
properties which could get more settings in the future, like F/stop,
shutter, resolutiongate, etc.

> - AFAIK, current exporters are supposed to export scene in way, in which
> rendrered "outside" blender image would match image rendenered in Blender
> (from composition POV, not materials/lights ofcourse). So if vertical FOV
> would be used by exporters, it should be exposed into RNA/UI in a way, which
> wouldn't confuse default workflow.

No, it is not only about exporting what is visible inside Blender
render output. It is about calculating the correct focal length based
on sensor height and vertical FOV. Some formats / apps use these
vertical values to calculate the focal length, and some use the
horizontal values. With the Channel format you need to know and set
the correct sensor height to get the focal length, so if you get a
camera from Nuke you need to be able to input the sensor height in
Blender to get the shared focal length. It is more automatic with FBX
but internally it uses the vertical FOV as a primary value, and again
these settings need to be exposed in Blender to import a FBX camera.

Wasn't these values exposed in RNA/UI by the sensor patch with
'data.sensor_height' and 'data.angle_y' ? What is this default
workflow it confuses? Modelling?

> Currently i'm working on a bit different stuff and i'd prefer stop you from
> telling "this feature is needed! No way if there'll be no this feature!". I
> know that it is needed, just i can't fork and work on several issues at the
> same time.

I'm terribly sorry to bother you, and love all the work you've done in
Blender. Just one last thing before I will leave you alone:

Please explain what was wrong with the sensor patch? It contained all
what we agree is needed. Again only related to the camera properties,
not match-moving. This "confusing workflow" makes no sense, sorry.

Best regards,

Ejner