FS-100 Picture Profiles

Haha, no problem. I did a lot of playing around with ProRes 422 and native MTS in Premiere to see if the source mattered. It did not in *that* app. Other apps may not be the same thing. That said, given I go in and out of After Effects A LOT, I do use H.264 coming in, but ALWAYS ProRes 422 coming out.
 
Rick Burnett said:
Haha, no problem. I did a lot of playing around with ProRes 422 and native MTS in Premiere to see if the source mattered. It did not in *that* app. Other apps may not be the same thing. That said, given I go in and out of After Effects A LOT, I do use H.264 coming in, but ALWAYS ProRes 422 coming out.
Click to expand...

I'm contemplating a switch to Premiere after FCPx debacle. At least to play with it. I am hearing many good things! Can it take in ProRes and output as well? Don't want to hijack this thread but curious.
 
You will need some form of Final Cut installed to export to ProRes, and you cannot export to ProRes on PC. You can read ProRes files just fine though.
 
Neex said:
You will need some form of Final Cut installed to export to ProRes, and you cannot export to ProRes on PC. You can read ProRes files just fine though.
Click to expand...

Yes, I forget this sometimes. I have FCP6 installed AND I am in OSX. The client I am working with now uses a PC and can ingest my ProRes, but edits everything on his end with DNxHD.

What do people on the PC side normally use with Premiere?
 
Uncompressed avi (mostly BlackMagic codec), DPX or Cineform.

HDnX is also nice but a bit a pain in the a... cause there is a bug in the software.
When you want to export HDnX from a Premiere timeline, the HDnX settings window that pops up is not fully expanded, so you can not
access all the buttons at least on my machine.

Frank
 
Rick Burnett said:
What do people on the PC side normally use with Premiere?
Click to expand...

Nothing, everything plays natively and smoothly in premiere so there is no need for the additional generational loss of compressed intermediate formats or significant additional storage requirements they represent. I always edit natively and use realtime GPU effects and all formats I've tried (XDCAMEX, AVCCAM, AVCHD, R3D) work seamlessly together on the same timeline. Usually I just export deliverables from the timeline, but if I need a final master I may use uncompressed or lagarith lossless or even high bitrate MP4 for less critical stuff.
 
nyvz said:
... there is no need for the additional generational loss of compressed intermediate formats or significant additional storage requirements they represent..
Click to expand...

I use both Cineform and Avid DNxHD. Cineform offers a non-destructive color correction module that works on their native 10 bit codec, but, it costs $$$. The DNxHD codec is free and also offers a 10bit variant. If you have an NLE capable of a 10bit workflow(not many can do that as of this time) my suggestion is that you give the 10 bit workflow a serious consideration. While 10 bit files are larger, if you do ANY CC or special FX, working in 10 bit makes noticeable improvements in image quality. Media Composer has a 10 bit workflow if you start with 10 bit DNxHD. Whatever NLE you pick, their internal FX have to be written for 10-bit.
 
The nomenclature is confusing, fersure. "32-bit" refers to 8 bits per channel, including alpha, or 8X4=32. If 10 bit color depth processing were to be referred to the same nomenclature, it would be called 40 bit. Beware advertising soft shoe.
 
cuervo said:
The nomenclature is confusing, fersure. "32-bit" refers to 8 bits per channel, including alpha, or 8X4=32. If 10 bit color depth processing were to be referred to the same nomenclature, it would be called 40 bit. Beware advertising soft shoe.
Click to expand...

I understand your confusion since that is usually how it works, but actually in this case, Adobe uses 32bits per channel, effectively 96bits per pixel. According to adobe, GPUs are designed to handle 32bit floating point values with the same speed as 8bit ones, and that seems to be the case. The power of GPU acceleration is quite impressive, so it is hard for me to imagine any modern NLE not taking advantage of it by this point.

Whats the point of a lossy conversion to an interpolated 10bit format from 8bit originals when the 8bit originals are going to be decompressed into 32bit uncompressed RGB internally for all operations anyway?
 
nyvz said:
Nothing, everything plays natively and smoothly in premiere so there is no need for the additional generational loss of compressed intermediate formats or significant additional storage requirements they represent. I always edit natively and use realtime GPU effects and all formats I've tried (XDCAMEX, AVCCAM, AVCHD, R3D) work seamlessly together on the same timeline. Usually I just export deliverables from the timeline, but if I need a final master I may use uncompressed or lagarith lossless or even high bitrate MP4 for less critical stuff.
Click to expand...

I don't mean on bringing in to editing, I mean on exporting to work with other programs once you have edited. I move files between different programs and that is where I use ProRes 422. It seems people use cineform, DNxHD, or uncompressed mostly.
 
nyvz said:
I understand your confusion since that is usually how it works, but actually in this case, Adobe uses 32bits per channel, effectively 96bits per pixel. According to adobe, GPUs are designed to handle 32bit floating point values with the same speed as 8bit ones, and that seems to be the case. The power of GPU acceleration is quite impressive, so it is hard for me to imagine any modern NLE not taking advantage of it by this point.

Whats the point of a lossy conversion to an interpolated 10bit format from 8bit originals when the 8bit originals are going to be decompressed into 32bit uncompressed RGB internally for all operations anyway?
Click to expand...

Whether GPU acceleration is actually working well is problemmatic....and arguable.

As for my workflow, I edit in Media Composer. It is very convenient for me to convert all my footage into DNxHD before I import into AMC. AMC uses DNxHD as its internal intermediate, so, transcoding ahead of time saves me on resources since I can transcode on a seperate machine rather than my edit station.

I think you may be somewhat confused between the difference between 32 bit floating point mathematical processing and 10 bit color depth. 32 bit FP math simply means the processor can use up to 32 significant digits for mathematical operations on 8 bit data. A color depth of 10 bit implies over a trillion different significant colors, as opposed to millions of colors as you have with 8 bit color depth. So, while floating point math improves color accuracy (by reducing round off error)and helps aliasing, it has no effect on the quantization effects that produce color banding, aka stairstepping. For that, you need true 10 bit color depth capability.

As a side note, I'll express my melancholy that it is no longer sufficient to be just a videographer or photographer in today's digitized reality. One must also be a mathematician and computer wizard to understand the nuances of this digital technology. :)

The REAL reason that 10-bit workflows have not been universally adopted is because there are no consumer 10 bit displays available in the marketplace, as of yet. You can look at any modern 8-bit display monitor and see stairstepping in subtle gradients, like sky, even if the workflow was 10 bit. Right now, 10 bit wokflow is almost an exercise, because 10-bit processing is coming in future releases of all NLE's.
 
Last edited:
Rick Burnett said:
It is not necessary in some apps to convert first to 10-bit on input, but, that depends on the internal representation of the video information. That said, what IS important is that you work in 10-bit and you output in 10-bit if you plan to do anything else with it. 10-bit through an entire process will reduce quantization errors in color correction and other image manipulation. ALWAYS work higher than your target bit depth. This is true for video AND audio. I just find it easier to convert EVERYTHING I do to ProRes 422 and work from there.
Click to expand...

Ah, so in Some apps it helps. In Premiere, which I'm used to, it does not matter. It has 32-bit float therefore if you shot in 8-bit, there is absolutely no reason to convert to 10 bit prior to editing. And I'm not talking about it playing better in the software, I'm talking about converting for quality differences. :)

But that's interesting to know...thanks!!
 
I've not seen any quality difference in Premiere or AE by transcoding to a different format first. I've tested it on multiple projects and feel confident in Premiere handling of the footage.
 
cuervo said:
It is very convenient for me to convert all my footage into DNxHD before I import into AMC. AMC uses DNxHD as its internal intermediate, so, transcoding ahead of time saves me on resources since I can transcode on a seperate machine rather than my edit station.

)
Click to expand...


You say it is convenient to convert your footage, instead of just dropping that stuff on the timeline and start working?

And you need a second machine for that?

I don´t get it.

I always thought, the most conviniant way of editing is, to just drop any footage, from any camera, on any drive, at any timeline and just start editing.

Frank
 
Postmaster said:
You say it is convenient to convert your footage, instead of just dropping that stuff on the timeline and start working?

And you need a second machine for that?

I don´t get it.

I always thought, the most convenient way of editing is, to just drop any footage, from any camera, on any drive, at any timeline and just start editing.

Frank
Click to expand...

I thought the same thing, I think many people are so used to ingesting their footage despite all the extra time and space it requires that they cannot get used to the idea that it could actually be faster and better quality to just drop footage into their NLE straight out of the original media folder structure.
 
You must log in or register to reply here.
Back
Top