NTSC square pixel 720×540 problem??? I’m a Brit!

[Robert Houghton]

Well for me that’s perfectly normal. We create our graphic plates in Photoshop at 720×540. When I set up for final output I make sure the end composition is 720×486 nonsquare pixel (.9 ratio). To make everything fit I just scale it down to 90% and hit the render button.

The only time that can mess you up is when dealing with video footage that has not been deinterlaced (via footage properties on import). If you deinterlace the video you should be good.

-Rob

Motion Graphics Animation
Professional & Freelance
Respond2

Opinions expressed above are not in any way connected to Respond2.

Personal website under construction 😉

[Jamesworsley]

ahhh i see.
I’ve also just read that people design in 648×486 square pixel comp before they drop into a 720×486 non-square comp, is this the same as designing in 720×540 square pixel and dropping it by 10% like you said?
Cheers
James

[Sam Moulton]

I’ve seen a test somewhere on the cow that showed two resolution charts, one scaled to 90% and one at 100%. One way you lost horizontal resolution, the other vertical. I’ve always worried about scan lines and how horrible they look when things are not right so I go for the 648 X 486 route most of the time for full screen artwork. I don’t like the horizontal edges to go soft. I really don’t think there’s much of a difference. I really don’t like scaling up D1 video to fit in 720 X 540 comps because of the scan line problems that can happen either.

I found the thread. You can see the difference by dragging the movie control between 90 and 100%. I made my decision by looking at the lines of text as well as the charts. However, most of the time my graphics are moving around so there’s really no point to worrying about the 90 or 100% scale for graphics rule anyway.

https://www.creativecow.net/forum/read_post.php?postid=107196265377204&forumid=50&archive=_2003

[Steve Roberts]

If you look in the COW tutorials for “Pixel Madness” or “Dr. Strangepixel” by Rick Gerard, some things may be cleared up.

[Jamesworsley]

Thanks for the help.

So to confirm if i do all my design in 720×540 sq. pixel in Photoshop, i’ll be ok dropping the 720×540 sq pixel After Effects comp into a 720×486 rectangle pixel comp and scaling down to 90% to render.
Just to let you know i’m only using photoshop layers, no video footage.

Thanks again, it’s great to be able to get advice from people!

[Sam Moulton]

I always use cmnd option shift H and not f… if there’s a mis match in scan lines or any other size, F will distort, h fits the layer to the width and preserves the pixel aspect ratio. I think it was Chris Meyer that gave me the idea. He said, at least I think it was him, that he never uses fit, only fit to width. You can get there from the the layer>transform window.

[Darby Edelen]

As someone who works regularly with video and photoshop layers for final output on NTSC displays I have to ask: what is the benefit of working in square pixels at all?

My entire workflow I use what I assumed was the ‘correct’ NTSC pixel aspect ratio of 0.9. In Photoshop my documents are always set to NTSC pixel aspect ratio… when I draw a circle in Photoshop and import it into After Effects everything is interpreted at 0.9 and nothing appears stretched (my circle is still a circle).

What is the purpose of preparing documents in square pixels if they will never be viewed in square pixels? It’s not a rhetorical question really, I really want to know if I’ve been wrong all this time and should rethink my workflow =/

[Darby Edelen]

[wuzelwazel] “what is the benefit of working in square pixels at all?”

I really wanted an answer to this… Nobody checking this thread anymore? (;

[Sandy Chase]

Side-by-side comparison I do prefer working in 648X486 because as far as I understand, going from 720X540, the pixels have to be reinterpreted/mapped to 90% scale; in other words, generating new pixels from an approximation of the original. Going from 648X486, the pixels are stretched, but each pixel keeps its original color/position, so there is less artifacting. It’s subtle, but on text and fine lines it can make a difference.

I do work in Photoshop with non-square pixels (especially helpful for HDV with its more extreme pixel stretching), but some programs, like Cheetah3D, I’m stuck in square pixels.

I guess it would be ideal if every program could accommodate different pixel shape views. Occasionally I run into the reverse problem; for example, exporting production stills from Photoshop and 3D graphics, where I designed everything for non-square video, but they need square pixels to print.