8 bit Vs 10 bit editing

[Jim N]

Edius is a 8 bit based editing system (ie 256 grey levels in each RGB channel)Some systems offer editing in 10 bits (ie 1024 grey levels per channel) Where is 10 bit video really important to use or is there real any advantage to editing in ten bits.

If Edius were a 10 bit system would it slow down?

Jim

[antonsvideo]

if you edit in order to produce for cinema release, yes, 10 bit is important

for the rest, not important at all, since you can't give anyone 10bit

[shueardm]

In compositing perhaps you can benefit but in editing video it's so neglegible.
Antons crorrect even Blu-ray is only 8 bit. BTW, Australia's only real studio endorsed Blu-ray authoring facilty (Motionlink) uses EDIUS HD which is 8 bit.

[Jerry]

Edius is a 8 bit based editing system (ie 256 grey levels in each RGB channel)Some systems offer editing in 10 bits (ie 1024 grey levels per channel) Where is 10 bit video really important to use or is there real any advantage to editing in ten bits.

The advantage is in graphic manipulation and compositing. The more robust the initial video the better the final quality will be after you have added or taken away from the video.
For most people, 8 bit is fine. I have found in gradient scenes, i.e. blue sky or underwater scenes, you can see the digital banding on the screen.
I have had very good luck with an 8bit uncompressed workflow in those situations.
Cineform has a 10bit compressed codec. It does work well.
Many of us here have requested a 10bit compressed HQ codec. With the sale of GV, I seriously doubt it will materialize.
In most cases 8bit will serve you well enough. It would be nice to have the option.

If Edius were a 10 bit system would it slow down?

If you are talking a 10bit compressed, yes it will slow down but not as much as a 10bit uncompressed system. It will also depend on your raid subsystem to handle it. If you can get two streams of uncompressed 8bit to work with a subsystem of 300 read and write on your raid, you will be doing fine.
That is why compressed formats are so popular.

[redgum]

Just about every broadcast studio in the world uses 10 or 12 bit uncompressed for the same reason you may use HDV instead of DV.
There is a considerable difference but it depends on your workflow and destination. Anyone working animation or Effects would think it mandatory.

[shueardm]

They may end up in 10 or 12 bit but along the way it's sometimes even passed around in SD (SVideo none the less) before it goes out to 10 bit HD. I've heard of stories from channel 7 etc.

[tingsern]

However, the editing system is only half the story. What about the cameras that acquire the video in the first place? It is useless to have 10 bits NLE if the input is only 8 bits. Today, I have not heard of an affordable professional level video camera that do 10 or 12 bits.

[cuervo]

It is useless to have 10 bits NLE if the input is only 8 bits. Today, I have not heard of an affordable professional level video camera that do 10 or 12 bits.

Not true! Read the post from "Jerry", above. And, the Sony EX1/3 will output 10-bit via the HD-SDI port. The real problem is field capture of the 10-bit 4:2:2 HD-SDI signal.

[GrassValley_KH]

This has always been the problem - cameras can output 10-bit...but recording to media always tends to be in 8-bit.

In the good old days, nothing was coming from a camera in 10-bit. That's how EDIUS was conceived - the camera's best friend. Now we live in an age of more than just camera-produced media (and better computers). And so, EDIUS will likely evolve..

[tingsern]

I am referring to video editing (NLE) only - not After Effects kind of work. YES - I agree that banding will occur with 8 bits only ... I have seen that myself as well. Especially if the gradient used is very steep - not enough colour shades there.

HD-SDI output - yes - that's about the only way to capture more than 8 bits from an affordable camera. But, as you pointed out, unless you do that in a studio condition, outdoor field capture is problematic.

[hanso]

GV infinity 10 bit

I'm using GV infinity Camcorder as has a 10 bit MJPEG2000 codec, I can see the fotage is a much better than it is from P2 HD, so the question is when is the material downconverted to 8 bit in Edius?
It seems like the fotage is played back in 10bit on the timeline.

Hanso

[GrassValley_KH]

When you apply any effect, or render (out) from the timeline, the results will be in 8-bit.

[PaulTV]

Flash XDR will record uncompressed 10 bit from the EX1/3 onto compact flash cards in the field - it's just a bit expensive! $6000 ish.

Paul

[Jerry]

I am referring to video editing (NLE) only - not After Effects kind of work. YES - I agree that banding will occur with 8 bits only ... I have seen that myself as well. Especially if the gradient used is very steep - not enough colour shades there.

HD-SDI output - yes - that's about the only way to capture more than 8 bits from an affordable camera. But, as you pointed out, unless you do that in a studio condition, outdoor field capture is problematic.

I think that we all would agree that if you are capturing from 3 1/3 inch chips compared to 3 1 inch chips you will see more color information....heck, more information period. This is where the HD SDI import to 10bit will matter.
Even in Cineform 10bit compressed codec you can see the color difference.

It all becomes choice. For me personally, if HQ had a 10bit compressed codec I would be using it more so than not.

[cybertrix]

Where is 10 bit video really important to use or is there real any advantage to editing in ten bits.

Well, seeing as I’m waiting for an HD render… here goes!

For an average edit which doesn't require much colour grading... 8bits is enough and you won't run into too many problems. If you are grading from an 8 bit source, and delivering an 8 bit master - hey - you may as well be seeing an 8 bit preview like the end user will see!! I constantly argue this with purists who usually who run systems named after a fruit work in 10bit just because it sounds better.

However... I've recently come to understand why 10bit rocks.

And there are 3 reasons for this:

1)))))

I've been using the Red camera quite a bit on music vids lately and keeping the edit workflow 10bit HD uncompressed. This is because the record company insists I use their edit suite and the DOP use their camera. To be honest, the whole Red thing is a little overhyped (in my opinion) IF … note this if… IF you are just using 16mm lenses in the windowed 2k mode. I don't know why ppl buy a Red then never use 35mm lenses. The GV Infinity would produce a better and less noisy image given the equivalent lens it's 3 chip sensor block (let's not go there though!!) Basically the Red shoots 12bit and leaves the image totally flat so all the contrast/colour adjustments/black crushing can be done in post. I always find that "black" is usually sits around a muddy 25% luma value and there is a ton of overshoot. Most cameras would "crush" this in camera before going to memory / tape (as the GV does with a 14bit DSP, then saves at 10bit). So I "convert" the RAW 12bit files to a "crushed" 10bit and basically pull the "muddy grey" blacks closer to 0 while pushing out the highlights out leaving a bit of overshoot just in case. This is then spits out 10 bit videos in horizontally cropped 1080p (down from 12bit, 2k).

Keeping a 10 bit editing workflow means that the end result can be passed to a grading package without all that "conforming" (which wastes time). Keeping a project in 10 bit space is vital if you are pushing and pulling lumas and colours later. If I were to pass my 10 bit files into Edius I'd get approx 7 bits of detail after the grade :-[ Not so much in a linear fashion, but after applying an "S" curve (to get a more filmic look) the mids would potentially be stepping 3 or 4 values. And you notice - big time. Especially on a flat face. It's still a bit noisey in 10bit, but you don't notice as much. In my view, if the camera op knows what they are doing, they can do some of this operation using the in-camera DSP at 14 bit and get a pretty solid image from the camera, even in 8 bit.

As a realist, I tried switching between 8bit / 10bit to see if it was all just a waste of time but there was a very noticable increase in banding and noisefloor in low contrast shots working in 8bit. Especially floor tiles, wooden doors and flat faces which you are trying to add contrast to.

I also looked at 10bit vs 12bit but couldn't spot a major difference once it was back on a digibetacam tape or an HD H264 compared to the 8 bit flow.


2)))))

Now for the other thing: When you are working with Rec 601 "broadcast safe" 8bit YUV footage - the luma levels are generally sit between 16 (black) and 235 (white). IF you expand this to RGB by pulling 16 to 0 and pushing 235 (ish) to 255 you are going to get banding. To see an approximation: Make a gradiant in photoshop from level 16 to 235 then do an "auto levels" on it. See? There are approx 36 "inbetween" values which have to be doubled up. In reality noise usually hides this. But by having a 10 bit source you can technically expand to RGB with much cleaner results.

Also - and here is one people often miss - going from HD 8bit YUV > Standard Def 10bit YUV can mean picking up additional color values. If we have the luma value 77 and 78 sitting next to each other in HD and combine them to make an SD pixel, we get 77.5. The .5 is lost when converted to 8 bit space. However this is retained in 10bit space as the value 310. So later on in the chain, if we expanded our 10bit YUV to 8bit RGB we'd be filling in those 36 "inbetween" values.

So, in short, my answer would be YES! There is a real advantage to editing in ten bits IF:

1) the source is above 8bit and you plan on doing any color correction later down the track. Especially when pulling up darker areas or expanding greys.

2) If you plan on finishing in expanded RGB at a lower resolution and wish to get the best possible downconversion from an HD source

and finally:

3) If you want to brag at a christmas function "well I work in 10 bits because it's just better". Indeed. But just don't make a claim like that to me, because I'll be questioning why ;-)

If Edius were a 10 bit system would it slow down?

Technically, I guess approx 60% - 70% of the speed for the most part. The main question is "would existing filters / effects work at 10 bit?". And the answer would be "not without re-coding". And I doubt the third party plus would work in 10 bit either. Converting a package from 8bit to 10Bit or 16bit is not trivial. It requires a 10bit workflow every step of the way - probably a complete re-engineering of the package. But fingers crossed.

Even if it were just a mode with the main reframing tools, colour tools, an HQ codec and an export option to move cleanly into CODEC's like BlackMagic 10bit (which is industry standard from what I've seen) or QT 10bit... then it'd certainly get Edius up a notch. People are pushing towards a higher bit depth - seeing as the display panels are starting to emerge. And it’s all the buzz with 30-bit sRGB now standard in Windows 7...


All this and I'm still only 87% rendered.