transcoding to similar format/codec as my native footage

O

offbeatbryce

Well-known member
Hello,

my camera films in ACVHD .mts files. My camera records in 28mbps max in 24 and 30 fps. I've tried exploring transcoding before but all the codecs/formats given to me built into premiere etc are DNXHD or Cineform which takes a 400MB file and turns it into a whopping gig. I would like to retain a similar file size and quality as the original. I film documentaries and often times don't need half the footage on my editing drive but as of now have to because I can't find a right format to transcode to.

I'm using a Windows 10 computer.
 
Others may give better advice, but when I only need smaller parts of long AVCHD clips I usually transcode to .mp4 at a higher bit rate of like 35+.
Much smaller size than DNxHD etc. Is .mp4 not an option in your Premiere? It is in mine, as well as .h264.

Seems to have worked pretty well so far over the last several years. The resulting clips allow me to save just the scenes I need without having to save the entire clip.
Although on a side note, if I think I may need it again, I actually save the original clips too, just in case
 
offbeatbryce said:
I've tried exploring transcoding before but all the codecs/formats given to me built into premiere etc are DNXHD or Cineform which takes a 400MB file and turns it into a whopping gig. I would like to retain a similar file size and quality as the original.
Click to expand...

Why do you want to transcode?

The file sizes of your originals are a compromise: actual frames of footage are removed and replaced by parts of frames and mathematical algorithms and predictions. When being played back, the restoration of the full frame takes "lots" of computational power. But you do save space this way, since the math is calculated in real time and not "baked into the file".

This realtime calculation is an extra burden for the computer to carry on top of everything else going on in an editing/coloring environment. Overhead. Stress.

In addition, at the time of shooting your footage, if you record to a long GOP format you will likely see artefacts in frames with lots of movement, like water or foliage in wind. You can't change this later, it's just a downside of these more economical codecs.

A good reason to transcode would be to move from inter frame compression (your current long GOP, with lots of math) to intra frame compression, like ProRes, Cineform or DNxHD. These latter formats unpack the math (decompress) and restore all the original frames. Now you have a proper "photo" for each of your 24-60 frames (typical) per second. These frames are also compressed, like jpegs, but all compression is WITHIN each frame and there is no more compression over a GROUP of frames. No motion artefacts (if you capture directly to a format like this). But now you see where the increased file size comes from: lots of extra, real frames. Also, these latter formats hold up very well over several generations of compression.

Original footage (1st gen) --> transcode (2nd gen) --> edit/color --> export (3rd gen)

The same can't be said for long GOP encoding. Each time you recompress to a format like h.264 there's a much bigger risk of introducing unwanted artefacts. This brings us back to your original suggestion. Hopefully, by now you can see that going from long GOP to long GOP with roughly the same data rate (maintaining size) would only introduce potential artefacts via generational degradation without bringing any other benefits.

Transcoding doesn't make your footage magically better. If your computer can edit the original files, just stick to that. If you have performance issues, do a proper transcode and delete the larger files when done.
 
Last edited:
AndreeOnline said:
Why do you want to transcode?

The file sizes of your originals are a compromise: actual frames of footage are removed and replaced by parts of frames and mathematical algorithms and predictions. When being played back, the restoration of the full frame takes "lots" of computational power. But you do save space this way, since the math is calculated in real time and not "baked into the file".

This realtime calculation is an extra burden for the computer to carry on top of everything else going on in an editing/coloring environment. Overhead. Stress.

In addition, at the time of shooting your footage, if you record to a long GOP format you will likely see artefacts in frames with lots of movement, like water or foliage in wind. You can't change this later, it's just a downside of these more economical codecs.

A good reason to transcode would be to move from inter frame compression (your current long GOP, with lots of math) to intra frame compression, like ProRes, Cineform or DNxHD. These latter formats unpack the math (decompress) and restore all the original frames. Now you have a proper "photo" for each of your 24-60 frames (typical) per second. These frames are also compressed, like jpegs, but all compression is WITHIN each frame and there is no more compression over a GROUP of frames. No motion artefacts (if you capture directly to a format like this). But now you see where the increased file size comes from: lots of extra, real frames. Also, these latter formats hold up very well over several generations of compression.

Original footage (1st gen) --> transcode (2nd gen) --> edit/color --> export (3rd gen)

The same can't be said for long GOP encoding. Each time you recompress to a format like h.264 there's a much bigger risk of introducing unwanted artefacts. This brings us back to your original suggestion. Hopefully, by now you can see that going from long GOP to long GOP with roughly the same data rate (maintaining size) would only introduce potential artefacts via generational degradation without bringing any other benefits.

Transcoding doesn't make your footage magically better. If your computer can edit the original files, just stick to that. If you have performance issues, do a proper transcode and delete the larger files when done.
Click to expand...



But transcoding to Cineform or dnxhd give bigger file sizes. I can’t eben store my footage on a 1TB ssd because when I transcoded to those formats my avchd footage from the sdxc card which was 1 GB turned into 1.5TB just from the transcode
 
offbeatbryce said:
I transcoded to those formats my avchd footage from the sdxc card which was 1 GB turned into 1.5TB just from the transcode
Click to expand...

That doesn't make sense. In your first post you say 400MB becomes 1GB (1000MB, simplified). And now you say 1GB becomes 1.5TB (1500 GB, simplified). That's going from 2.5x to 1500x.

Your footage is 28 Mbps. ProRes 422 would be (aprox) 117 Mbps, 4.2x, for HD and 471 Mbps for UHD. Other intermediate formats would be similar to ProRes.

Why are you transcoding?
 
AndreeOnline said:
That doesn't make sense. In your first post you say 400MB becomes 1GB (1000MB, simplified). And now you say 1GB becomes 1.5TB (1500 GB, simplified). That's going from 2.5x to 1500x.

Your footage is 28 Mbps. ProRes 422 would be (aprox) 117 Mbps, 4.2x, for HD and 471 Mbps for UHD. Other intermediate formats would be similar to ProRes.


Why are you transcoding?
Click to expand...

If you read my original post I want to transcode only footage I’ll need. I film documentaries and only end up using half the footage I film. This is what documentary filmmakers do. I generally film several interviews with the same person and wind up editing together what I want but then I have all the interviews from the sdxc card which I currently transfer to my 1TB SSD.

Instead I only want the interviews I’ll use on my SSD not all of them.

My entire sdxc card is 1 GB. Before I was giving an example of 400MB.

I am on windows 10. Don’t have pro res.

Want I want to do is to take my avchd footage select in and out points then export the in and out points to some other format so it’s only want I’ll actually be using in my premiere project.

I also want to label the clip names because currently my avchd footage on the sdxc card is labeled with random numbers when my camera records.
 
AndreeOnline said:
I appreciate gaining deeper insight into the intricacies of film making. We're all here to learn.



It's a tricky situation. I hope you find a solution..........
Click to expand...

Is there anyway to trim the .mts files before I import? Say for example I had taken video of a birthday party and grandma was blowing out the cake but there something before that I didn't want at all no matter what. Could I trim the .mts file so that only the cake blowing was present? This would reduce the file size of the .mts file right?

Another thing I read about transcoding was that color correction was better but I'm not seeing how that can be when my camera films in 8 bit mode and all the cineform codecs are 10 or 12 bit. Wouldn't i just be wasting space by transcoding? I don't see how having a larger file will all of the sudden make the color correction better or look better.

my friend who edits only transcodes certain effects to burn them into the clip if he has a greenscreen effect and doesn't want premiere using up the effect resources anymore
 
It looks like you need trimming, not transcoding. However, I'd be very careful with discarding footage. When editing, often one would use footage that initially looked unnecessary.

For editing (trimming, splitting, joining) .mts files without any transcoding and without any lose of quality (it keeps the original CODEC, only trim, split and join it), there is a freeware application, TSSniper: https://www.videohelp.com/software/TSSniper . It works well with .ts and .m2ts files, it may work also with .mts files.

This is the latest version, 0.80.0.600 test, form May 24, 2009, though it works very well even now.
 
You must log in or register to reply here.
Back
Top