1080i HDV to 720P or 480P or 270P

Question

1080i HDV to 720P or 480P or 270P

I am filming in HDV 1080i. My final products are going to be delivered in 720p, 480p and 270p (internet).

I film concerts so I thought I could use the digital zoom or SCALE effect within FCP 5.1.4 and I would not lose any noticeable definition because my final output is going to be in a smaller size.

I am now confused how to figure out how much I can scale images that start in 1080i without losing quality (resolution) for a final output in:

(a) 1280 x 720p
(b) DVD NTSC 480p
(c) 480 x 270P (internet)

Can someone explain a simple formula so I can figure out how to do the calculation. I know I could do it by eye, but I want a simple math/science answer that makes sense to me and preferably a simple explanation.

Thanks

Dual 2.5 Ghz PowerPC G5, Mac OS X (10.4.9), 4.5 GB RAM, 2TB (4 drives) RAID; Aja Kona Lh Capture Card

Posted on Jun 7, 2007 12:37 AM

Reply

Answer 1

Jun 7, 2007 11:34 PM in response to Matthew Thomas4

You keep bringing up issues i have solved...ie field reversal. I know you trying to be helpful, but that's not helping me. I don't have a problem with field reversal as I go to H264 to a progessive scan and have had no problems there.

I am paying with the scale of my rock concert zooming in more on some scenes and less on others and transitions between the footage to make it look like I had filmed it with 5 cameras.

There is an EXACT formula to at least get real close to the actual result. I know there is an art to look and feel and sharpening and softening and color correction, etc...and I am not asking about that.

When I put an HDV 1080i video into a 480 x 270 space without scaling it looks fine...BUT I can digitally zoom in or SCALE within FCP on the original image before converting it...if I scale by 110 or 120 or 130% no matter what you can never see any degradation in the final output of the final output is so much less and the the resolution of the raw footage is so much greater. It works well. Now if I try to scale 500% you will see massive pixelation because the resolution of the original footage is just not there.

Same with still pictures that are 5 or 6 or 8 megapizel shots...you can zoom in 200% on most of them and if the original looked good the zoomed in will look fine even in HD...because the picture is so much higher resolution. And even 300% in near standard def format. You cannot do this with consumer shot video as the detail is just not there to begin with.

If you do that with video you usually see massive pixelation. Of course a 4K Red camera you could probably do this 200% or more and still have great images!

There is a simple mathematical way to figure out when you start to lose resolution purely based on the resolution. This is my STARTING ground I am trying to figure out. I tweak it from there based off on artisitc needs and other minor factors some of which you mentioned. I just can't figure out the math behind it and I want to to help me for all different outputs...I want to understand the concept and formula that describes it.

Do you understand what I am trying to do?

Reply

Answer 2

Jun 7, 2007 8:48 PM in response to Stefan Boyland

You can use compressor and create a preset that uses Custom Frame controls. Set the Resize Filter to Best. Set the Deinterlace to Motion Adaptive and the Rate Conversion to High Quality Motion Compensated. Use these settings to create a High Quality Master of your movie in each resolution (preferably uncompressed or in ProRes). Then go to bed. It will take all night to build even short clips. In the morning you might like like what you see. It is very very slow so I would suggest you use this technique to make "Master" versions at your targe size and frame rate. Then compress them into your delivery formats from these masters. This way you won't have to run the frame controls on your various compressed versions.

Perhaps you could reverse telecine the footage in Cinema Tools to 1080P then just downscale it to 720P, etc. I wouldn't recommend this since I tried it in FCP 6 last night and it was a nightmare due to the way FCP messes with pulldown patterns.

Reply

Answer 3

Jun 7, 2007 9:39 PM in response to Matthew Thomas4

I know how to make movies with compressor to any size I want.

My question is within final cut pro when I am edting my camera angles together I am trying to create more looks and feels because I only shot with 2 cams. hI am trying to use artificial zooms and flipping the video to give it a feel like I have 4 or 5 cams going. My exact question is how much can I SCALE my 1080i HDV concert video knowing ahead of time that my final output will be:

1280 x 720
720 x 405
480 x 270

I am trying not to lose noticeable quality so when I output to smaller sizes I know I can use this "digital zoom" or scale technique to give me more options with the video.

I'm looking for the formula of how to figure it out...and possibly the answer to these 3 resolutions-

PS (I am pretty sure if I scale something about 130% that fits in 1280 x 720 without loss (again fcp crops out the 1440 x 1080 and then compressor resizes it to my custom setting so I get to see a zoomed in shot of part of my video frame), 160% for 720 x 405 and maybe 480 x 270 maybe 200%. But how does one figure out the exact number for this?

Reply

Answer 4

Jun 7, 2007 10:11 PM in response to Stefan Boyland

Sorry. If you want to create an arbitrary formula for the sake of... whatever, I'm not interested in participating in this thread.

I am interested in helping you make your footage match but I doubt this should be done by a formula.

A 720P plasma or LCD HD monitor would be your best indicator of loss in this case. Turn down sharpening on the monitor before judging detail.

The best scaling ratio is not always the same because some (most) cameras over-sharpen footage. Sharpened footage doesn't scale up well (at best 30 percent). Smoother unsharpened footage scales much better since you won't have as many artifacts.

Field reversal will bite you no matter how you to scale with 1080i. Converting it to 1080P before editing could be a solution, but I wouldn't recommend it mid-project. Do it before starting to edit.

Loss of resolution is in my opinion second to field reversal in importance if you plan to make the footage match. Pulldown removal on HDTV's is not reliable if you keep mixing up the cadence and field orders. You may have to deinterlace each clip before scaling up or down which will hurt your detail.

Reply

Answer 5

Jun 7, 2007 11:45 PM in response to Stefan Boyland

goodbye. it was nice trying.

Reply

Answer 6

Jun 7, 2007 11:46 PM in response to Matthew Thomas4

Thx for trying-

Reply

Answer 7

Jun 7, 2007 11:48 PM in response to Stefan Boyland

OK. One last question. How have you solved the field reversal problem? Other than removal, I'm not aware of a way to solve this while maintaining the detail you expect.

Reply

Answer 8

Jun 8, 2007 12:00 AM in response to Matthew Thomas4

I edit in HDV the original footage...I edit multiple SEQUENCES which I label as 1080 or 720 or 405 or 270 based on the number of vertical lines of resolution. The first number is always 16:9 ratio so 1080 is 1920 x 1080 H264 file...720 is 1280 x 720 H264 file..again for final output.

When I put the HDV file into compressor I set have custom presets I've used now and sometime tinker with for bit rate, resolution, audio, codec, etc..and I use ones I built based off of tons of my own testing. I do not DEINTERLACE the footage because in H264 Quicktime movies the output looks the same as interlaced and I think that is because the H264 codec in Quicktime viewed on a computer is always progressive. The output files no matter the size always look spot on as compared to the original HDV file, except you cannot see the interlacing which I like better.

When I downsize from 1080 to 720 or 405 or 270 it crushes all those vertical lines together anyway, so the detail I lose is not pixelation but rather resolution because my final output is in a much smaller format as to the original editable footage.

I edit and break up the video without changing anything in HDV...the native format. I scale and do all of that natively...that is HDV. The footage going in is the footage coming out...it is not until the very very end when I am complete that I then output using compressor to H264.

I've done this 50 times in the last 2 years and never have a problem, except now I am making more copies for the web & email that is why the 480 x 270 resolution and I'm curious about how much zoom on the original footage I can use before I see it on the output. And again that depends on what my output is in each case.

I imagine you either didn't understand my work flow or why it is I am trying to do what I am doing. I've done multi cam editing for concerts now with up to 8 cams...and it works real well this well. Just not sure just how far I can get that "digital zooming" or "scaling" in...often I am too far from the performers, so I do want to give the illusion that my cam is much closer to the performer. I can approximate for now and it works fine, but I'd like an exact formula for my own understanding purely again for resolution.

Hope that explains what I am getting at.

Thx for your patience-

Reply