best way to de-interlace

[Mike Albertini]

ps… i do have after effects cs5 at my disposal if thats the best option…

[Vincent Rosati]

I’m sure someone else could chime in with more up to date info, but Premiere’s default deinterlacing, as you indicated, would probably be the fastest, though it is the lowest quality. It does what I call a ‘field doubling’ method. You essentially lose half of your vertical resolution.

Another method you can do in premiere is to duplicate your clip, so the identical clip is on a layer exactly above the original. Then on the lower clip, use premieres default Deinterlace/Lower fields. Then on the upper clip Deinterlace/Upper fields. Then set the upper clip to 50% opacity. Less information is lost with this method, but if you have a lot of motion there will be problems, sometimes referred to as ghosting.

I’m unsure of the latest default effects available in PP or AE, but for the best quality you would want a deinterlacing method that isolates the visible combing in an interlaced frame, and deinterlaces only the portions of the frame that need it – like Fields Kit.

A big however – If you’re ultimately just going to reduce the size of the video for web, I’d use the fastest/premiere’s default deinterlace. Render the file. Then put the deinterlaced render into a new project, where you would perform the resize and render your final version.
I’m sure there are a dozen other workflows you could use to do this, but this is just a suggestion.

Vince

*Please remember to Rate our replies or check Solution if solved. If you get a good idea from the post, consider clicking the Kudos option.

[Mike Albertini]

Hi Vincent,

Thanks for the reply…I could try duplicating all the clips but that will take a while considering its 40 mins long…I’m thinking about buying fieldskit…just downloaded a demo…what do you think of it?

I’ve also just applied magic bullet frames to the sequence…waiting for it to render…i had this already…have you tried this – what do you think of it?…

btw i also have a section where i speed up a clip to 1900% and also a sequence where i slowed down the clips to 50%…i did this just using the default speed changer in ppro
…ideally i’d use twixtor or kronos for this but again cant get hold of them at the moment

Think I’ll re-do those using timewarp in after effects…is that the best method in after effects?..i’ve read its ‘based’ on kronos…

Lastly – h.264…is there an alternative to this codec for small file sizes for the web…I’ve read about gamma issues and when i open my project in mpeg streamclip or vlc i am sometimes getting different color variations depending on what settings i chose on export…it looks ok in quicktime though…

Thanks,

Mike

[Chris Tompkins]

Duplicating clips only takes as long as it takes to hit copy and paste.

However, You are best to deinterlace AFTER you export.
Export to AME and THERE you can deinterlace to the web delivery.

Chris Tompkins
Video Atlanta LLC

[Mike Albertini]

it does indeed, but not exactly an efficient way of de-interlace a project is it?…

Anyway, in terms of de-interlacing after you export…if i press command>E… and want to export in h.264…what settings do i use to make a good de-interlaced clip?…Do i just change the field order to progressive?…Where are the settings to control the de-interlace procedure?

And if you say its better do de-interlace after you export, do you think fieldskit deinterlacer is not worth getting?…

Thanks,

Mike

[Ht Davis]

I like your method to use the doubling method, but I don’t like using the 50% drop. We’re talking FIELDS, which are placed every other line, with nothing in between. The awesome part? When you overlay the frames, they should line right up. This also provides some new ideas on combining multiple low res cams into large res.

I’ve seen a guy test this and get decent results. It takes serious craziness and a hunger for lunacy and really huge frames, but it seems to work fine.
You need multiple of the same camera. IF you want to shoot progressive frames out of interlaced cams, you can get a 60p frame rate and a large amount of frame. by lining them up just right with just over half overlap horizontal and only 1\3 overlap vertical, you can get progressive large frame with a lot of cheap small cameras in an array that are synchronized to some frame-accurate timecode. Horizontally, every other camera should be flipped upside down. For every 2 fields you’ll get 2 frames. If you process it the same way as you said, but overlay each cam properly, and erase the areas of distortion, you’ll end up with a longer frame. You just have to process it correctly, and do so in after effects. The upside-down cams capture the lower field of their neighbors, hence the overlay of just over 1\2. With the timecodes all syncronized somehow, you can flip the upside down over in AE and stitch it right up. With 2 cameras, you’ll get about 50% of the frame size of the camera (50% of 1080p kind of sucks), and at 3 cams, you’ll get about 66-75% full frame. But you add the 2\3 to 3\4 frame size across for every camera you add horizontally after 3. The outside area will be interlaced, and need to be cut off. If you do this for several vertical rows, and then try to stitch them, you’ll gain 2\3 vertical extension for each step up, and you’ll be stitching progressive to progressive. Do the same in each vertical row that you just did. They should match in direction, camera for camera, upside down to upside down. They need a max of 1\3 overlay with the frame below and should be vertically aligned. When you stitch progressive to progressive, you’ll be overlaying full frames one on top of another, and there will be an area of distortion you’ll have to remove from the top. You could also just output the JPEGS, and have a script grab from each vertical row, send to a photoshop droplet that stitches them, and drops them to a folder. You could watch the folder in AME, and have it render afterward to a video file. At 3 cameras horizontal, you should have nearly a full frame, about 2\3. This means you can make your own 4k video camera out of about 6×6. The horizontals should be very close together. No more than a few feet apart. In fact, it may even be best to keep them only a few inches from one another. With 6 cameras about 6 inches apart, you’ll span 3-4.5 ft with tripods alone; and with a straightline rig, you can place other cameras on top of those. I’d put them no more than 2ft apart top to top. You’ll reach 12ft in a hurry, and you’re best off with another tripod behind the others with an angle stabilizer holding your whole rig. You wont be able to move it, but it works.
It’s a cheap and dirty way of getting huge sensors.