5D and CS5. Should I transcode before grading?

Adam J McKay

Veteran
Hey,

I am in pre production on a feature to be shot on a 5D and 7D. We intend on using the cinestyle profile for production. Now that CS5 supports DSLRs natively, is it still necassary to transcode before grading. Would there be anything gained by transcoding to cineform? Or does Adobe do the same job with the native files? I would test this myself, but I dont want to dish out the money for cineform first and ask qustions later.

Thanks,
 
I don't shoot on DSLR or use Cineform, but I've been working with all sorts of video codecs natively in Premiere for years and don't know what you would gain from transcoding to another format. It's not going to get any better than what you're pulling off the camera as far as color information or image quality are concerned.
 
I don't shoot on DSLR or use Cineform, but I've been working with all sorts of video codecs natively in Premiere for years and don't know what you would gain from transcoding to another format. It's not going to get any better than what you're pulling off the camera as far as color information or image quality are concerned.

+1
The entire transcode step is an artifact from when people were doing tape ingest, or a workaround if you don't have a up-to-spec workstation for CS5.
 
transcode canon 5d/7d footage to mxf will use much less CPU
smaller file size with pretty much no visual difference
 
CS5.5 only kind of works with native files. It will work with the native h264 files while it background process converts them to another codec (I forget what the default is...). So it actually is not a native work flow as I understand it. This is simply a way to remove processing from the user-end.

And for DSLRs at least, transcoding is not just an artifact left over from tapes. The camera records in h264, which is an end-user, processor intensive, condensed codec. This codec is not designed for editing workflows, specifically color-correction and grading. You should always use an intermediate codec built for the purpose of modifying files and will be less processor intensive (and generally larger file). This will have less impact on your processors, and should be capable of doing more types of manipulation without rendering, and also will have better quality (less banding, artifacting, larger dynamic range, etc).
 
CS5.5 only kind of works with native files. It will work with the native h264 files while it background process converts them to another codec (I forget what the default is...)
No, Premiere Pro (any version) does not convert any files in the background to another codec. Where have you heard/read that? :)

/Roger
 
I swear I remember reading that before, and from a legitimate source I thought... I am frantically going thru my history to find this misinformation.

But why would it natively work in such a processor intensive codec? It just sounds like murder on your computer. I understand from a user end wanting to avoid transcoding, but unless you have a top-notch computer, I am going to be surprised by how well it works to render h.264 files when modifying. I really need to download a demo of CS5 to see what it's all about...
 
But why would it natively work in such a processor intensive codec? It just sounds like murder on your computer. I understand from a user end wanting to avoid transcoding, but unless you have a top-notch computer, I am going to be surprised by how well it works to render h.264 files when modifying. I really need to download a demo of CS5 to see what it's all about...

CS5 is a 64 bit app that is multi core aware. My understanding is the other popular NLE is a 32 app that can only use one core, thus the transcoding.
 
64-bit doesn't guarantee your computer will haul ass. I do definitely look forward to FCPX which is rebuilt as 64-bit.

But my understanding of the h.264 codec is that it does poorly in anything other than playback, as it was written and built as purely an end-user codec with blocks and motion-oriented encoding that was not intended to be used for anything other than as final distribution. That being said, it's not that it can't be used in NLE, it's just that why would you want to put that extra load on everything to make something perform in ways it was never intended. There are dozens of very good intermediate codecs designed to be used in NLE and with better encoding for grading and color-correction.
 
No, Premiere Pro (any version) does not convert any files in the background to another codec. Where have you heard/read that? :)

/Roger

Adobe guru guy at adobe .com in one of the cs5 features videos say it converts it to a 444 realtime. (long haired white guy, Jason?)

@natums When I first bought CS5, I did my own little check to see if it played H264 realtime. I got 8 layers of 1080P , all with some kind of effect (color, blur, speed, movement, contrast etc) 8 layers played realtime, all at once. (with mercury playback enabled) (Magic Bullet put a damper on even one layer and would not play realtime)

That being said, If I was shooting a feature or something that DIDN'T have a quick turnaround, I would use cineform. (I have the cheapo ver)
 
Adobe guru guy at adobe .com in one of the cs5 features videos say it converts it to a 444 realtime. (long haired white guy, Jason?)
Yes, but it does not convert it to another codec and write it to disc. It does the upconvert during playback/encoding on the fly, but it does not put it into another codec on the HDD. Premiere Pro always works with and from the native files. :)

Point is, you will always only have your native files on your HDD's. :)

(Yes, it is Jason in the video.)
 
I actually have the same question/issue except with the 7D... I have edited only with the native H264 codec at 1080p in Premiere CS5 and colored with Magic Bullet Looks, and it looks fine after exporting back to H264 for youtube. But after uploading to youtube, the colors and artifacts are horrible. The images become horrible.

This makes me wonder whether I should have converted the files to Cineform in order to increase the depth to 10bit, but would this have made the video looks cleaner in its final recompression on youtube? Any input would be greatly appreciated.
 
Now that CS5 supports DSLRs natively, is it still necassary to transcode before grading. Would there be anything gained by transcoding to cineform?

Stay in your capture codec, or transcode to an editing codec? Like almost anything you can ask, it depends. In this case, it largely depends now what kind of color grading you are going to do, how much, and what your final output will be.

For example, if you are just going for DVD release and aren't going to do extensive color grading, then you might want to stay with the native codec.

On the other hand, the higher you go for final release (blu-ray, 2k or 4k digital release, film output) the more important your color grading, and the more extensive it might be.

The more you do, the more headroom you need. If you native chroma subsampling is 4:2:0 like AVCHD is, you might indeed want to transcode to a 4:2:2 space, or even a 4:4:4 space. If you don't do much grading or effects work, this won't gain you much. But if you do, this will give you some room to work in.

The other aspect of moving from a capture codec to an editing codec is that you'll trade compression for speed. In other words, you'll uncompress your capture footage some, and gain some speed in editing. Your editor will thank you.

So... it all depends. On what you need, and what you want.
 
For heavy color grading, youtube export, I got horrible artifacts while editing with native codec. But I had the same issue on FCP with ProRes for the 7D again. And I've tried and tested every export setting. With very clean footage with proper lighting and settings. Youtube just slaughters the quality, but others with 7D videos on youtube don't seem to have as much of a problem. I even tried with light grading and still had heavy artifacting.
 
Back
Top