Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

The Other Big Mistake

M

Mark L. Pederson

REDuser Sponsor
Joined
Dec 28, 2006
Messages
7,145
Reaction score
1
Points
0
Age
59
Website
www.offhollywoodny.com
I've said it before and I will say it again - 10bit LOG is not the future of post production.

I'm not going to say you can't make pretty pictures color grading 10-bit LOG DPX files. I'm pretty sure every film nominated for best cinematography at this years academy awards used this old tired and true pipeline.

But ... things change.

Change is good. Technology raises the bar. You can fight it - or you can step up.

Sensors, cameras, monitors and projectors are all pushing past 10-bit processing.

10-bit LOG is a pretty good way to pack a lot of info into a container. But the containers - and info - are bigger and better now. You want to process your images and transform them with precision and fidelity.

If you look at the cost of CPU, GPU and hard drives costs for a 8 bit HD pipeline in 2003 (the year I started OFFHOLLYWOOD with Aldey) vs. a 4K, 16-bit pipeline in 2010/2011 - guess which one is cheaper?

Checkers of Chess. Your choice.

RAW. LIN. 16-bit. OpenEXR or 16-bit DPX format for mastering.

Just say "no thanks" to 10-bit LOG.
 
Would love to see an example...but my monitor can't display it. We need better displays in general as well. :}
 
Thank you Mark! I'm tired of saying this around here and nobody listens. Check out my signature. Enough already.

Thank you, thank you. Thank you again Mark.
 
Did a bit of exr vs dpx testing.

And then all vfx is done as exr.

Don't have an option for that for colors... yet.

Testing RAW online on a shortie, now... again...

Until now better results from RGB files. That may change.

Lack of full metacontroll has been the biggest reason for holding off on RAW online...

And the goodies RED release in their own software...

I don't neccesarily agree that log is dead, though. But 10 bit log is getting long in its teeth...
I think it is more likely that the 10 bit dpx's may have had their time in a year or two...
 
Yes, 10bit log will die soon, time to spend some money in post...
 
Shane

Shane Betts said:
Thank you Mark! I'm tired of saying this around here and nobody listens. Check out my signature. Enough already.

Thank you, thank you. Thank you again Mark.
Click to expand...

I prefer - " If you always do what you always did, then you'll always get what you always got!"

Change is good!

Great post Mark - thanks

Steve
 
Change is good but it will take a long time to get implemented.

Look at DEMs vs. GOPs with many political fights right now...
 
As long as the transition is tested and verified, it is good.

Sometimes one swap good options for less good, just to change.

But that's a more philosophical thing.

As to 10 bit DPX, their days are counted.
 
While you have a 35mm print as your main output, I think we'll have to stay based on a 10-bit log workflow. Just put an 3D lut over to monitor, send the files directly to print, and another 3D lut to make the digital deliverables from it.
The other thing I like from a log image is that it can contain more dynamic range than a normal image in relation to a normal screen. Maybe a good replacement will be exr, but it's not clear to me if color grading software can handle it.

Congratulations Mark!
 
Steve Johnson said:
Shane
I prefer - " If you always do what you always did, then you'll always get what you always got!"
Steve
Click to expand...

:laugh:

But sometimes people are "happy" with what they always got. "Better the devil you know ..." and all that. For me, when the only argument someone can present amounts to, "But we've always done it that way", I think it's time for them to saddle up for change.

Then again, if the old way actually works better ...
 
My 117 cinematography awards from all over this world.. says, you don't need RAW, 35mm, 8bit, 10bit or 16bit to have great images.

It all comes down to content and execution of each shot, beautifully strung together as either a well told story or aspirational images.
Whether you shoot 8mm, super 8, DV, 16mm, Digibeta, HD, EOS 5, 35mm, RED.. what ever.... it does not matter.

I saw some of the best images of my entire life shot on one of these crappy first generation nikon HD DSLR video cameras, But then I keep thinking, it's not the camera at all, it's the human behind it. The Photographer was world renowned and has only just discovered video. (I quickly steered him towards the RED site)

Guns don't shoot people, people shoot people.. And old saying that flows into photography. The camera is useless, unless a crafts(person) picks it up and uses it correctly. It has nothing to do with latitude or the brand of camera at all.

I've won cine awards on almost every one of these crazy formats over the last 15 years. I'm telling you that you seriously don't need RAW to have a great image. But it does help if you want to grade deeper.. but thats about that.

That all said.:smilielol5: I personally can't wait to have an edit time line that I can drop R3ds directly into it without the need of rendering, like I can do with my 2k REDLOG PRO RES files. Ooooh I relish the day.

I do agree though. 10bit is just a temporary substitute for when we can all edit, grade and master direct from R3D's in 4k from a FCP time line with minimal rendering.
Come on APPLE pull your fucking finger out!
 
This is not the "other big mistake."

The big mistakes are: not enough light and having a crappy story.

Most people are satisfied completely with 8 bit displays. Right now we don't even have 10 bit displays in common use. In fact a lot of people are using 6 bit computer displays with dithering- and that is the best display they own!

(That's FRC or "frame rate control" dithering, not spatial dithering. Basically the dithered LCD's do their dithering in time rather than in space- so you don't see fixed patterns. It's what I talk about throughout this post.)

You can deliver all the results they could desire with a 10 bit linear RGB process.

Cinema display is more demanding, because its big enough for people to see the difference and the display system itself has more precision and DR. People are completely satisfied with 10 bit linear RGB display in that application, and 10 bit LOG gives a professional the room they need to deliver a quality 10 bit linear result.

That much is acknowledged in the OP.

Until I can see a deliverable result that is flawed in a way an "average audience" is likely to notice, and that the flaw is something that can not be corrected in 10 bit Log space, I have to differ this does not number amongst the big mistakes.

In fact I'd argue it isn't a mistake at all, just a compromise.

I agree 16 bit linear calculation is the post standard I'd like to see next.

Before we see that I think we have to improve our diffused display technology. Basically we should bump every display up a notch. Until then even 10 bit linear is overkill.

8 bit dithered displays need to replace the 6 bit dithered displays. (In other words, we should have a dirt cheap system that approaches 10 bit.)

10 bit dithered displays should replace the 8 bit displays dominant in the market.

12 bit dithered displays should take over the high end consumer and low end professional market served by both 8 bit dithered and 10 bit displays today. (Eizo and HP Dreamcolor monitors)

12 bit dithered and 16 bit displays should be available and replace the 10 bit displays used in professional applications today.

Plasma could fit some, but not all, of these needs, but in practice we'll have to wait for a next generation display technology.

I also believe that the market story of plasma displays should be considered.

Many here would agree plasma is the best and cheapest display technology on the market today. A plasma TV has the best color fidelity and response time of any commonly available consumer TV in the marketplace.

Consumers routinely pass plasma HDTVs by for much lower spec higher cost LCD displays. Guess what- a lot of those consumers are our clients- and they are all part of our audiences.

Why would they do that? What does that tell you about the market for pristine high bit depth workflows?

Let's not optimize for an edge case prematurely. If we wait a bit, until audiences become accustomed to looking at high quality displays, then the edge cases will become more common

I differ with Mark Toia slightly. All my 3 crappy awards have been for footage shot on XL-1 MiniDV and Canon 7D. I've never won for RED, DVCPRO HD, HDCAM, XDCAM or film. Beyond a certain minimum bar, it isn't about the format.

We do need 8 bit color to deliver good images. 6 bit, even dithered, looks like crap. So 8 bit becomes the minimum standard. Not a point of consideration, since even the "Barbie Video Girl" shoots 8 bit.

I'd argue that there are no HD formats that substantially limit the ability to do acceptable work, and even award winning work.
 
My awards are won in dv...
And I still stand by the production that collected a bit of recognition.

There are two sides to this argument.
The one Alexander and Mark T represents is at least as valid as the OP.

We are going exr and possibly raw where that pays off in addition to sticking to the 10 bit log thing, where that's the practical choice. Because we have found that to be valuable in side by side workflow tests.

But bits stops and resolution is a marketibg thingy, too. I am very much of the "Get out and shoot and post with whatever tool is practical and available for you" philosphy.

Really... I have said it a couple of times, but I still love the M image, even after having worked with the MX for a while.

Great technology doesn't immediately become obsolete just because there is better technology out.

r1M is still great technology and it works really well. So does 10 bit log post...


Cheers

G
 
gunleik is correct. there are two extremes being represented, who, ultimately probably have the same goals. of course story is the crucial factor for anything, without it, you have nothing. however, that is beside the point in this discussion, that is given. lighting, also, should be pristine, done well on set - that is a given, and also beside the point.

this is a technical discussion of post-production, not pre-production or production itself. in post, we want to work with the images we have to get the best out of them to TELL THE STORY. this involves tweaking and manipulating the visual data. the more room we have to tell the story through images, the better. is it possible to shoot something straight on a dv tape and get the point across... yes. but we are talking about top of the line, professional practices, and we're talking about ideal workflows in order to get the most out of our visuals. and there will be a day soon that is not today, surely we know this? can you really tell me that you can get the same quality image out of a dv camera? that would be silly... computers are very helpful tools, and used in the right way, you can definitely do things you could never do "in-camera" (and yes sometimes there are things that would be better if we didn't touch a computer). so when we're talking about digital cinematography, it's given that a computer is going to be used to perfect the images, and that computer is going to be much more effective than the chip in the camera. it helps to have that ability. it's not just a get into post and have a happy accident, it's purposefully planning for post (the three P's of Production, just made em up).

so everybody's right in a way, but let's not put up hedges, and let's definitely pursue a more perfect experience (for ourselves and the viewers).

story is important.

in visual storytelling... visuals are important too.
 
Totally agree with Paul in every way. Everyone is right, but not everyone is talking about the same thing.
 
16 bits or death

16 bits or death

I contend that the Cineon log curve typically used in 10 bit DPX log files is poorly matched to the linear image data from digital sensors. IMHO the road to great image quality at the point of exhibition (display tech being a separate but very important issue), particularly as digital projection supplants 35mm film in theatrical settings, is higher bit depths in addition to resolution, etc.

Cheers - #19
 
Good interview from Mark's post-super that echos these comments:

"DPs and producers are forming impressions of production technology, but the impressions are based on a round peg of digital cinema being wedged into the square hole of legacy post."

http://www.studiodaily.com/blog/?p=5057

Like shooting digitally, but the post-house using tape-based ingest and workflow, trying to milk every last dime out of those tape decks...
 
I've found 16 bit REDlogFilm DPX to be a very good compromise. If there are systems out there that give real-time performance (without cacheing first) working with EXR files, I haven't used one. 16 bit log DPX gives the performance of working with an integer file format, but doesn't constrain you to the 0-1 range, and doesn't run up against the granularity limit of 10 bits.
 
You must log in or register to reply here.
Back
Top