Rendering at HD

[3DDave]

As long as time permits, I always render at 1280 x 720 for the following reasons. I wouldn't call this the best HD resolution, but it is a start in that direction.

1. The Pixel Ratio is 1:1
2. I can deliver in either 4:3 or 16:9
3. The render is cleaner since more pixels are evaluated and rendered.
4. When reducing the frame size from HD to SD you still retain a cleaner image compared to a SD render.

[bedrock]

I see -

I THINK I dont understand what you mean by the pixel aspect being 1:1 - I'm familiar with the DV aspect of .9, and I generally never render at that... I'm always rendering at a full 1 pixel aspect (for aspect preservation?). What are the advantages of rendering at 1 over .9? I'm confused regarding the distortion and how that effects post production or broadcast.

I never thought of being able to deliver at 16:9 AND 4:3... permitting a "pan & scan" version of the film. That's great.

Excellent points. Thanks!

[3DDave]

1:1 is a pixel ratio of 1

1:1.3333 is a pixel ratio of .9

[Mauritius]

I THINK I dont understand what you mean by the pixel aspect being 1:1 - I'm familiar with the DV aspect of .9, and I generally never render at that... I'm always rendering at a full 1 pixel aspect (for aspect preservation?). What are the advantages of rendering at 1 over .9? I'm confused regarding the distortion and how that effects post production or broadcast. If that is true then combined with your statement above that you render at 720xX this breaks down to: all the images your ever rendered for broadcast had the wrong aspect. :)

NTSC/D1 is 720x486 at a pixel aspect of ~0.899 and a frame aspect of ~1.481
PAL/D1 is 720x576 at a pixel aspect of ~1.066 and a frame aspect of 1.25
When displayed finally on the TV however, both formats have a frame aspect of 4:3 aka 1.333!
The pixel aspect needed thus is the frame aspect of the TV divided by the frame aspect of the image:

NTSC/D1: 720/486~=1.481, 1.333/1.481~=0.899
PAL/D1: 720/576=1.25, 1.333/1.25~=1.066

If you e.g. want to render PAL at a pixel aspect of 1:1, you'd have to render an image 768 pixels wide (namely 720x1.066). This image before being written out to a PAL/D1 will have to be squeezed horizontally to 720 pixels (again giving the pixels the correct aspect of ~1.066).
On the TV screen, the image will later be displayed in a 4:3 aspect -- correcting for the squeezing. You
can think of a TV screen implicitly applying a non-uniform scaling to your image for which you have to compensate beforehand to cancel out its effect. If you don't do this, your image will end up on the TV distorted -- circles will be ellipses etc. etc. In a PAL image the effect is only roughly 7% but in NTSC its 10% -- enough to be noticeable.

Cheers,

Moritz

[playmesumch00ns]

Even in PAL you start to wonder why your characters' heads are all squishy. :)

[Andrew W]

Even in PAL you start to wonder why your characters' heads are all squishy. :)
And why CG characters that are supposed to playing football appear to be playing rugby...

A

[Mauritius]

Yeah, I should've put the "only" in quotation marks.
I guess that most TDs fall in the category of people that can spot immeditaly if only 2.5D motion blur was used, textures are not filtered correctly or the pixel aspect is wrong by 0.01%. ;)

Cheers,

.mm

[playmesumch00ns]

Yeah but don't get me started on motion-blur-policing, pixel-f*cking 2d supervisors

[Mauritius]

Yeah but don't get me started on motion-blur-policing, pixel-f*cking 2d supervisors
Now you got me hooked -- what about these bl**dy 2d supervisors? :)

.mm

[rendermaniac]

As a 3D TD I just tell them "Don't worry, we'll fix it in the comp..." That starts worrying 2D people for some reason ;)

Simon

PS I am only joking.....

[3DDave]

If that is true then combined with your statement above that you render at 720xX this breaks down to: all the images your ever rendered for broadcast had the wrong aspect. :)

NTSC/D1 is 720x486 at a pixel aspect of ~0.899 and a frame aspect of ~1.481
PAL/D1 is 720x576 at a pixel aspect of ~1.066 and a frame aspect of 1.25
When displayed finally on the TV however, both formats have a frame aspect of 4:3 aka 1.333!
The pixel aspect needed thus is the frame aspect of the TV divided by the frame aspect of the image:

NTSC/D1: 720/486~=1.481, 1.333/1.481~=0.899
PAL/D1: 720/576=1.25, 1.333/1.25~=1.066

If you e.g. want to render PAL at a pixel aspect of 1:1, you'd have to render an image 768 pixels wide (namely 720x1.066). This image before being written out to a PAL/D1 will have to be squeezed horizontally to 720 pixels (again giving the pixels the correct aspect of ~1.066).
On the TV screen, the image will later be displayed in a 4:3 aspect -- correcting for the squeezing. You
can think of a TV screen implicitly applying a non-uniform scaling to your image for which you have to compensate beforehand to cancel out its effect. If you don't do this, your image will end up on the TV distorted -- circles will be ellipses etc. etc. In a PAL image the effect is only roughly 7% but in NTSC its 10% -- enough to be noticeable.

Cheers,

Moritz

I my case, after I complete the compositing process I add transform node to save out to the required frame size and pixel ratio. I use Digital Fusion 4 for all compositing.

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.