[Michael Goldberg]

Happy New Year,
I just wanted to update this for anyone interested.
I think it was an H264 issue I was seeing.
Since the original footage was .MTS, I decided to try to output to the same format (or at least comparable). I used MPEG2 Blue Ray set to transport stream, when outputting this from a 1080i sequence with everything set at default UF – I got the exact same footage as the input, at the same file size.
I went back to H264, and found that if I set the bitrate up to the maximum (300), I got pretty much the same result. Only difference was the file size was huge. At lower bit rates to H264, I was still getting a little bit of artifacts (probably acceptable). I decided to output progressive, and that also gave me very good results and similar file size to the original. All in all, a good test.
The original problems seemed to be due to too small of a bit rate with H264. Thanks for the suggestions though.

Mike

[Michael Goldberg]

Thanks Jeff,
I am probably just looking at this too closely.
a Few questions for you –
When you edit natively at 1080i, I assume you’re footage is being interpreted as Upper Field first, and you’re timeline is also set to Upper Field First?

When export progressive, effectively you’re discarding a field, or interpreting the second field with some type of blending, correct?
Are you not noticeing a slight quality loss?

This has given me the best results, but I do see a slight degradation (which might just be inevitable if I want to go out Progressive), when I export this way. It’s very slight I admit, but was just trying to get the best results.

Thanks,
Mike

[Michael Goldberg]

Thanks Jerry,
I can get the footage to display nicely in Premiere. Generally just displaying 1 field rather than both gives me the correct look. My concern is that I’m slightly degrading the footage once I export.
What I was hoping was to not deinterlace on export, and have the footage look exactly the same when I play the final H264 in WMP or in VLC. I’m seeing a slight degradation.
When I deinterlace, and then export as progressive, this looks pretty good, but I can notice a degradation in quality which is inevitable since the second field is either being thrown away, or interpolated at that point. I’ll keep experimenting though.

Mike

[Michael Goldberg]

Hi Vince,
Yep the frame rates are consistent. I’ve tried interpreting the footage as progressive, making the timeline progressive, and rendering progressive, still noticeable lines. I’ve tried interpreting as UF with timeline as UF etc. The bottom line is I’m seeing more noticeable interlacing lines in every option.
If anyone wants to take a look at the original shot, I’ve uploaded it here:

View Link: https://clientshare.ccfv.com
Project Name: MG Temp2
Username: gold2
Password: goldtemp2
filename: 00980.MTS

Note: For best results, download file to your local hard drive.

[Michael Goldberg]

Thanks Todd,
I just loaded the update, and same problem exists.
I guess I could just convert the files to a different format, Just trying to avoid all the extra work.

Mike

[Michael Goldberg]

Hi,
I can’t speak much for C4D’s tracker, as I don’t use it. I use syntheyes for tracking, and highly recommend it. This type of scene should be no problem to track. When You get a reasonable track in any tracker, it should be able to export locations of various tracked objects. Once you have these, you can correctly position your 3D objects and render these out. For something like this, I would probably create a mask to put a CG element behind something in the video. You can position a layer at a tracked location (for example if you have a tracker locator at a chair), and create a mask for a foreground object. This layer will act as a track matte for the CG layer that goes behind. You could ofcourse create 3D geometry to replace objects in the video as well. There are more than 1 way to skin a cat. One more object would be to use mocha which is built into AE. Mocha can allow you to create planes that would also hold mask layers to mask objects out.

Mike

[Michael Goldberg]

Hi Adam,
Yeah, I get the camera distances and how they work. I guess where I’m getting confused is that everything works as expected if I check Use gradients in the Post effect. If I uncheck Use gradients, it ignores the Rear and Front distances, and only uses the Focus Distance. Sometimes this gives me a nicer looking blur. However, when rendering a depth multipass, this does not match the blur if use gradients is unchecked. not sure why? I’m guessing if use gradients is off, it calculates the front and rear based on the furthest objects in your scene. I guess I’ll try and set my distances further to match that.

MG

[Michael Goldberg]

How do you do that?
My original render had settings for front and rear blur, but the depth map didn’t match. When I turned the front and rear off, it didn’t render any depth? Are you rendering a separate map using a separate scene file?

Mike

[Michael Goldberg]

Hi,
For anyone interested, I thought I’d follow up on this topic.
We can’t use Team Render right now because it doesn’t work with stereoscopic cameras. So I was back to figuring out the color shift between V13 Net renders of V15 files.
It’s actually pretty simple –
In V15, Specular and Specular Color were combined into 1 channel.
When I opened the V15 file in V13, I saw that the Noise shader that was limiting the specular in V15 was loaded into the specular color in V13. This channel was turned off. So, when I turned that on in V13, the renders looked identical. I guess I’ve got a workaround.

Mike

[Michael Goldberg]

Hi,
Yeah, I hear you. I’ve been using this method for about a year since we started using V15. I had some issues with VPN and Team Render. I did some tests yesterday, and I’m trying to eliminate other factors. For example. I was rendering to H264, and when I rendered to Animation compressor, there was a definite difference (or should I say the difference between that render and the local render was less). I’ll do some more tests, and see if I can’t find anything else out.

Mike