Re: 8-bit vs 10-bit calculations and math...

The advantage of converting to 10-bit will only be apparent with certain coloring effects, for example maybe something like working with a gradient or adding a smoke effect. These actions would be better off applied in 10-bit than in 8-bit, even if it is downsampled later in the pipeline back to 8-bit.

[I]"The advantage of converting to 10-bit will only be apparent with certain coloring effects, for example maybe something like working with a gradient or adding a smoke effect. These actions would be better off applied in 10-bit than in 8-bit, even if it is downsampled later in the pipeline back to 8-bit."[/I]

Indeed and that would hold true IF Vegas had a 10 bit pipeline but it doesn't. The only choices are 8 bit or 32 bit float.

I believe some NLEs do have a 10 bit pipeline as an option which probably explains why we hear of these ideas. They are not applicable to Vegas.

Bob.

That discussion centers mainly around the Premiere fanbase, who blog a lot about theoretical advantages, but never quite get around to demonstrating it. 90% urban myth, 10% fact in my opinion, same thing that has made the rounds about every 2-3 years.

[b]TANSTAAFL[/B]

"I believe some NLEs do have a 10 bit pipeline as an option which probably explains why we hear of these ideas. They are not applicable to Vegas."

There are 2 different parameters being discussed - one is the actual bit depth of the media file, the other is the bit depth that the software lives in. A 10-bit media file cannot be processed to its full potential if Vegas is in 8-bit mode. But if Vegas is in 32-bit float, then 10-bits are easily processed. There is ample headroom in that case. I routinely import 8-bit mts into vegas then convert to 4.2.2. 10bit MXF to play with.

------------------------------------
"That discussion centers mainly around the Premiere fanbase, who blog a lot about theoretical advantages, but never quite get around to demonstrating it. 90% urban myth, 10% fact in my opinion, same thing that has made the rounds about every 2-3 years."

I do concede that in my amateur set up where my monitor is not even 10-bit, I can't tell you that there is a perceptible difference when playing back output media on my TV. But my PC has lots of muscle to flex so I don't lose anything by working within a bigger mathematical space. In terms of math there is a benefit to convert 8-10bit for coloring, but if it is only noticeable to pixel peepers then why bother?

[I]" I routinely import 8-bit mts into vegas then convert to 4.2.2. 10bit MXF to play with."[/I]

As far as I'm aware MXF is only 8 bit, even the 4:2:2 variant.

To the best of my knowledge the only codec native to Vegas that's 10 bit is Sony YUV. Be prepared for some monster file sizes.

Bob.

To the best of my knowledge the only codec native to Vegas that's 10 bit is Sony YUV.

I believe HDCAM SR and SR lite are all 10-bit. The 880 format being 4:4:4.
XAVC Intra is 10-bit 4:2:2 in Vegas.

Vegas puts both of the above in an MXF container.

Serena, this thread got somewhat sidetracked not that I have a problem with that. So let me address your comments. I have just downloaded the Sony Cyber-shot RX10 operator’s manual and reviewed the specs. I also downloaded an original MTS file uploaded by Gordon Laing to his Vimeo site and played around with it in Vegas Pro 12 to see how far I could push it before it fell apart. I was amazed and delighted how solid it remained, and how noise-free – way better than files off the BMPCC graded in Resolve. So more and more I am gaining confidence that the RX10 will deliver as promised and when it comes to video it will in all likelihood deliver much superior footage, is my early assessment.

But here’s the thing; it also shoots photographs and every day more and more are appearing on the Internet, and they are DSLR quality. And all that in one camera not much bigger than the BMPCC. The combining of such a complete video set with DSLR photo quality in such a small walkabout camera may not be of interest to many but for me it is huge, and I surely did not see this coming.

So, seems to me the RX10 just rendered my BMPCC obsolete. And therefore yes, I shall probably sell it as I cannot see me ever using it again, especially as I simply dislike the experience; that is the actual shooting with it for all the reasons I have articulated. Persevering will not change anything; I know what it is and what it isn’t, and most of all it will never make me a happy shooter no more than it will shoot DSLR quality stills. Of course your experience may well differ as your needs, desires and expectations are undoubtedly quite different. It’s not a right and wrong thing but simple a function of needs.

Your comment about my questioning the specs of the BMPCC I must briefly address. I’m not frustrated, well not in regards cameras. When I read comments that imply data, such as ‘feature film style 13 stops of dynamic range’, then I immediately ask how was that determined, and where is the supporting data? I can’t help myself, and that springs from an eternal deep-seated skepticism and incredulity towards unsupported pseudo-scientific slogans, especially those coming from businesses notorious for catering to the pretentious. I believe questioning to be a natural outcome of a healthy mind, to be encouraged, rather than a symptom of frustration. I would go so far as to suggest not to question is not only unhealthy but a symptom of severe oppression, insecurity, and highly unfortunate. But again, that’s just my opinion in response, and I’m not pushing anything here :-)

And as things have unfolded the 13 stops claim, for me, is moot.

Cheers.

[I]"I believe HDCAM SR and SR lite are all 10-bit. The 880 format being 4:4:4.
XAVC Intra is 10-bit 4:2:2 in Vegas."[/I]

You're right, HDCAM SR is certainly 10 bit. That must have been added in V12 which I don't currently have installed on any of my PCs hence I didn't see that option.


Even so there's still no mathematical advantage to re-encoding 8 bit to 10 bit.
There is a disadvantage to doing this though. The footage is being recompressed in the process and that brings with it a very real risk of degrading image quality.

In summary, yes, there's a theoretical advantage to processing 8 bit footage within a fatter pipeline, that avoids the potential for multiple rounding off errors to accumulate especially if there's a lot of FXs being chained. There would even be an advantage to render the output of that pipeline to a codec that's 10 bit.

Simply re-encoding 8 bit footage to 10 bit does no good and may do harm.

Bob.

http://provideocoalition.com/pvcexclusive/story/the-blackmagic-pocket-cinema-camera-reviewProVideo Coalition review[/link]

http://provideocoalition.com/pvcexclusive/story/blackmagic-cinema-camera-in-the-fieldBMPCC accessories[/link]

There is another issue with Vegas and certain codecs that has. Not yet been brought up. Vegas expands the levels from 16-235 to 0-255 on footage from certain cameras. It's actually a VFW problem, but Vegas inherits it because it uses VFW.

This happens with MOV footage from both my Nikon DSLR and my Panasonic GH3. It does not happen with AVCHD footage from my GH3. It also happens with footage from my GoPro. I also suspect it happens with Canon DSLR footage but I have no direct experience with Canon DSLRs.

I would strongly recommend converting footage from any camera that shoots MOV format just for this reason. It will save you some hard levels clipping for those occasions where there is detail in the 0-16 or 235-255 range.

Yes, Canon DSLR needs the leveling too.
A little surprised that it does not affect avchd from your GH3. Are you getting 16-235 or 16-255?
May be the Sony decoder is taking care of that for you.

If I shoot AVCHD I get 16-235. If I shoot MOV with everything else the same, in Vegas I will see 0-255. If I shoot so there is a corner of a bright window in the frame, in AVCHD it will be mostly 16-235 with a bit of action in the 235-255. The same thing in MOV will give me the same shape stretched out to 0-255. Running this through a cRGB to sRGB correction will look like the AVCHD histogram except that the bright window stuff will be hard clipped.

Playing the same MOV clips in various media players:

In WMP the MOV and AVCHD clips will look the same. In VLC (which also uses VFW) the MOV clip will look like cRGB levels. In Adobe Photoshop (and I'm assuming Premiere Pro) the clips will look the same. In Vegas the MOV levels look like cRGB. Basically, in anything that uses VFW, the MOV levels are 0-255. In anything that uses DirectShow, the levels are primarily 16-235. Yes, you can correct, but detail in the highlights and shadows will be hard clipped.

Converting to Cineform with their program will save the detail in the highlights and shadows that will be lost if a VFW program like Vegas stretches the level ranges.

This is not something I've read, it's something that I know through extensive first hand testing. I posted a bunch of these tests in a previous discussion about Mercalli SAL. Mercalli SAL reads the video at 16-235 levels then writes this to a format that Vegas also reads as 16-235. That is why the Mercalli levels appear to be different. They aare actually correct. Converting to Cineform using their conversion software bakes in this same correction, at least if you are using Vegas. In the process you will be saving a tiny bit of occasional detail in the 0-15 and 235-255 shadows and highlights range.

VLC doesn't handle decoding directly through system codecs, including vfw. Everything goes through the ffmpeg libs, and the resulting player levels may be different than either native system vfw (Vegas) or directshow (Windoes Media). I run across this from time to time in my tests.

"Converting to Cineform with their program will save the detail in the highlights and shadows that will be lost if a VFW program like Vegas stretches the level ranges."

......So, may I assume you are converting to Cineform from within Vegas using Cineform as a plug-in?.....That being the case, are you rendering with the cRGB to sRB levels filter on the output? If not, do you then insert this filter on the resultant project output?

From everything else I have been reading here, there is almost no instance where inserting the 16-235 filter on the output is not a good thing.

Having started working with conversion to Cineform a couple of months ago, I can say that, dealing with the OP's original question, the resultant ability to use the Cineform color correction tools has massively improved my ability to deal with the problematic HDV footage in my current project. Some of this has to be from additional headroom in the files, other from the color tools themselves.

If you are converting MOV format to Cineform, you are far better off using the Cineform software to do the conversion. Using Vegas will bake in the highlights and shadows clipping.

"If you are converting MOV format to Cineform, you are far better off using the Cineform software to do the conversion. Using Vegas will bake in the highlights and shadows clipping"

So, you are saying you do the Cineform conversion outside of Vegas, rather than using Cineform as a plugin for rendering from Vegas? What is your workflow?

I have a Canon G15 and have been bugged by levels in that applying cRGB to sRGB filters has not seemed to maintain detail in the shadows or light areas. I read Laurence's comments in the other thread but have stayed out of it for lack of understanding.

Tonight, I downloaded a trial of the GoPro app and converted one of the Canon G15 MOV files to Cineform. Based on the scopes, Vegas decodes the MOV levels 0-255 and the Cineform in an AVI wrapper more or less 16-235.

MOV

Cineform

If I'm all wet, let me know.

That is alarming.
I wonder if the problem is how Vegas is interacting with Quicktime?

Someone [I]really[/I] needs to bring this to SCS's attention.

Bob.

The problem is mostly With MOV formats but not completely. I also notice the levels being stretched with GoPro footage but in that case. The levels aren't neatly in the 16-235 range. It makes me suspect that the problem is not with QuickTime.

>Tonight, I downloaded a trial of the GoPro app and converted one of the Canon G15 MOV files to Cineform. Based on the scopes, Vegas decodes the MOV levels 0-255 and the Cineform in an AVI wrapper more or less 16-235.

Yes, that's exactly my experience. Did you notice that while the Cineform converted footage is almost completely in the 16-235 range in the scopes, there are little bits that occasionally go above 235 or below 16? Those are the bits where if you use the MOV footage directly and compensate for the expanded levels with a cRGB to sRGB correction where you are going to see clipping.

More analysis:

I placed the original MOV on the Vegas Pro timeline with the Cineform AVI below it. With no filters I rendered to the Sony AVC Blu-ray codec at ~22 mbps. The difference is apparently significant.

MOV-Cineform.m2ts
MOV-Cineform.ts


I added a cRGB to sRGB filter to the MOV at the track level and rendered the difference to the same render template.

MOV plus cRGB to SRGB - Cineform.m2ts
MOV plus cRGB to SRGB - Cineform.ts