ACES and ACEScc Intro and Tutorial for Vegas Pro

Comments

astar wrote on 5/11/2015, 12:18 PM
@Balazer - nice work on the tutorial.

I think PC hardware needs better Floating Point speed to handle EXR and linear format in a best/full fashion. Also, there are some advances in the latest versions of EXR that include compression to deal with file size issues. Vegas seems to only support EXR 1.0 which is uncompressed only.


If you work in EXR, and use the integrated build proxies, switching between Good/Best and Draft will choose the proxy in Draft. Switching to Good/Best will display the full res file.


How do you determine which plugins are ACES compatible, full range? Would you be able to list them?


If you are not shooting in sLOG, is the "default" color space the 709 IDT?


For RAW files like .CR and .NEF, will vegas convert these files properly into linear space? How does the IDT factor in with those formats?
astar wrote on 5/11/2015, 1:14 PM
@ wwjd ACES at the moment is more like a warehouse that you can do work in and not have information clipped. I do not believe there is a camera made today that can capture the ACES color space. 601, 709, UHD(8-10bit current flavors), DCI, and Rec 2020 all fit with room to spare inside ACES.

The fact that Vegas can support editing in ACES, means that you are rec.2020 ready by means of an IDT. The source footage still determines how high you can fly though.


balazer wrote on 5/11/2015, 3:54 PM
wwjd, a Rec.2020 ODT is available. I haven't added it to my config yet because I don't have the means to test it. By shooting with a camera that has a camera-specific IDT and editing in ACES or ACEScc, you are Rec.2020-ready. In theory you can render the same project for Rec.709 or Rec.2020 delivery just by changing the output color space. In practice it's good to proof your output on a Rec.2020 monitor before delivering in Rec.2020. A camera that has a wide gamut and high dynamic range will take better advantage of Rec.2020.

astar, I've determimed which filters are compatible with ACES just by testing them. Very few work well, so if you really want to work with filters, ACEScc is going to be easier than ACES. ** Rec.709 is the standard color space for HDTV. If you don't have a camera-specific IDT, use Rec.709. ** I don't believe Vegas can read any raw formats. If you have raw files, they'll need to be processed first, for example using Lightroom or dcraw. On my web page I have an experimental workflow for CinemaDNG raw files. The same process should work for .NEF files. Another option would be to export from Lightroom using a Rec.709 ICC profile, and then use the Rec.709 still image file color space in Vegas.
wwjd wrote on 5/11/2015, 4:13 PM
if you stretch 8-bit color spec, and/or contrast, and/or resolution from smaller 709 to 2020, what happens to the inbetweens created by Vegas? In ACES?

The quick, unscientific tests I did shows Vegas creating "inbetween" colors, instead of copying nearby pixel color info - keep in mind I barely know what I am talking about - so, isn't it possible to expand 709 into 2020, getting more tween colors by the process (even though they are not the original colors from a 2020 camera sensor) letting Vegas algorothms fill in the color spaces? ....it's just a theory I have. Like cranking up the bass and treble to "expand" or enhance an audio recording....

balazer wrote on 5/11/2015, 6:02 PM
Vegas does not interpolate colors when transforming between color spaces. The IDT maps the camera's color values into the high precision floating point working space. That's a virtually lossless operation, since 32-bit floating point has much more precision than any camera color space. On output, the working color space values are transformed into the output color space. That's a lossy operation, since the output color space has 8 or 10 or 16-bit precision. The end-to-end result is that your camera's precision is preserved, to the extent possible given the inevitable rounding errors that happen when you transform from high precision to lower precision. The output rounding errors are added to whatever rounding errors were already present in your camera's recording. You can minimize the cumulative output rounding errors by starting with a higher precision source. Ideally if you're delivering for 8 bits you'd shoot in 10 bits or more. But 8-bit input to 8-bit output is just fine in practice most of the time.

Vegas's many different forms of processing can certainly create color values in between those which can be produced by your camera.

Rec.2020 specifies high precision encoding, which could enable high dynamic range video. But it does not specify how HDR video would be encoded. There are basically no standards right now for HDR video, aside from the proprietary Dolby Vision. The Blu-Ray Disc Association is supposed to be making an open spec for HDR, which may or may not be the same as what the UHD Alliance is working on. For now we're going to just have to wait and see how the HDR standards will shake out. If you shoot in a scene-referred format and edit in ACES or ACEScc, you'll be ready for whatever standards come.
audio2u wrote on 5/11/2015, 9:03 PM
@balazer,
I'm reading this thread with interest, even though a lot of it goes straight over my head.
I'm an audio guy, and a keen photographer.
My photographic bent gives me an understanding of bit depth (I always shoot RAW, never jpeg), and to a lesser extent, an understanding of colour spaces.
I get the notion of shooting RAW gives me greater control of levels, and the ability to remap the luminance values of pixels prior to exporting to jpeg (for blogs, facebook etc) and so forth.
And although I have no intention of becoming a video-shooter/editor-for-hire, I am always keen to improve the quality of the home videos I produce of our family vacations. Yeah, I'm a geek! :)
I'm not shooting with the kind of cameras I hear mentioned in this thread. I am however, familiar with them (the A7S would be my next camera if my A-850 DSLR got stolen tomorrow). My home videos are shot on a (high-end domestic?) JVC GZ-HD6. 3CCD, 120GB internal HDD, blah blah blah. I really like it.
It stores its files with a .TOD extension, but they are essentially M2TS files, and I've found that I can simply rename the file extension to .m2ts and Vegas will play with them happily. Premiere Pro on the other hand would not, but that's no longer a concern, as I much prefer Vegas.
Occasionally, I find the GZ-HD6 has a tendency to over-expose its footage. Not massively, but enough to cook the highlights. Perhaps I need to take more control over the shooting mode and not allow it to operate in full auto all the time.
Sorry if I'm rambling. I'll get to my questions....
1) What bit depth are those m2ts files? I'm guessing 8 bit, but would be keen to have that confirmed.
2) Would attempting to implement your workflow (as described in this thread) give me any likelihood of recapturing those blown highlights?
Thanks in advance for your input.
john_dennis wrote on 5/11/2015, 9:43 PM

"1) What bit depth are those m2ts files? I'm guessing 8 bit, but would be keen to have that confirmed."

Download a utility like Mediainfo and see for yourself. Uncheck and don't accept any crapware.

I'm going to stay out of #2, but my theory is that if the camera sensor clipped, then no.

balazer wrote on 5/11/2015, 10:00 PM
audio2u, m2ts files are 8-bit. Nothing can recover blown highlights in the footage you've already recorded. But an ACEScc workflow can help prevent blown highlights in a couple of ways.

One way is to use the camera's auto exposure but to dial the exposure compensation in your camera towards the negative by 2/3 of a stop or so. You'd then boost the image back up to the proper brightness in the ACEScc workflow. In those instances when your camera misjudged the exposure and made the image too bright, you'll need less brightening in the workflow.

The other way would be to shoot with fully manual exposure control (shutter speed, aperture, and ISO). It's quite difficult to shoot with manual exposure and get the exposure right all the time, so you'll compensate for the exposure in the workflow, making the picture brighter or darker until it looks right. It would be your responsibility to make sure the images don't blow out in the camera. But the workflow gives you some exposure latitude, so you can afford to be a bit off in the camera and still correct it in the workflow. Since there's no camera-specific IDT for that camera, you would use the generic Rec.709 input color space, and render to Rec.709 output. The Rec.709 input color space won't be a perfect match for your camera's native color space, so the range of corrections you'll be able to make without obvious inaccuracies will be limited, perhaps up to a stop in either direction. But that might be enough. If you aim for an exposure about half a stop below the correct exposure, it means you could be wrong by up to half a stop in either direction and still correct it pretty well in the workflow without getting too much noise in the shadows or blown highlights.

Of course using auto exposure is easier, and using manual exposure gives you more control. Basically in ACEScc, Vegas's brightness filter becomes just like the exposure compensation slider in Lightroom.
audio2u wrote on 5/11/2015, 10:07 PM
Thanks John, I'll check out the link you provided!
And thanks also Balazer! I shall do some experimenting over the next week or so. We're heading off on a 3 week trip to Borneo (Sarawak and Sabah) at the end of June, and I want to shoot the best footage I can (given my skills and gear).
set wrote on 5/12/2015, 4:01 AM
Hi Jacob Balazer,

Well, at least I have more acceptable workflow on handling S-Log2, as ACES is still a new thing to me.
Previously? I had difficulty understanding the ACES, and in the end I just applied standard color curve and saturate in 8-bit color space. I knew the workflow 'not quite right' that before...

Since the equipment for wider color gamut seems still rare (as so far I can understand it), especially for my customers, I can only use it for more contrast scenes, so only very-very few shots, just grade it first with ACES workflow, then continue doing cutting in 8-bit default color space.

Of course there's a limit of 'contrast' level especially after I test on the last scene, where the camera pointing my computer and window. it shows the problem.
I (or most common user too?) may surely think S-Log2 as a 'magic' solution on balancing the extreme contrast scene, but it turns out not so magical, and noise is also easy to appear, that's the 'stopping factor' for having extreme adjustments. Probably caused by ISO3200 minimum for S-Log2 ?

For the window shot, I also test the underexposed one, to prevent a clipped highlight on window, so the interior was too dark, but in the end It wasn't work too, the scene was too noisy.

https://dl.dropboxusercontent.com/u/29342552/VegasPosts/ACESScene05-ACEScc-PP7_SLog2_MoreUnderSource.jpg

Maybe I need to search for downloadable high end camera footages samples like F3 or F5/F55 (or probably FS7) to find out.

As for ACEScc workflow, I had one strange result when comes to Saturation adjustment. I had white random 'dead pixels'-like on dark area.
Here's one sample where I boost Saturate to 100 (just for experiment purpose only) :
https://dl.dropboxusercontent.com/u/29342552/VegasPosts/ACEScc%20SLog2%20FullSaturateTest.mp4 - it shows up on preview monitor.


It's a shame, though, that the exposure meter only reads up to +2.0. You just need to dial an exposure of +2.0 or below, and then make the adjustment in the camera to reach +3.0.
How to push +3 ?


Set
balazer wrote on 5/12/2015, 2:07 PM
As for ACEScc workflow, I had one strange result when comes to Saturation adjustment. I had white random 'dead pixels'-like on dark area.
It looks to me like an artifact in the operation of the saturation filter. I would try a different filter. Vegas has several that can do saturation

Your shadows are noisy because the shot was underexposed. You need an exposure around +2 or +3 in S-Log2 on the a7S. S-Log2 does not magically let the camera handle extreme contrast scenes or give you tons of highlight range. What it does do is give you more exposure latitude than you'd have otherwise. Aim for an exposure of +3 if you want to minimize shadow noise. Aim for an exposure of +1 if you want to maximize highlight range. You've got a good 2 stops or so of exposure latitude, which is only a stop better than a typical camera.

One way to set exposure on the a7S in S-Log2 is to use the exposure meter. Just watch the exposure meter while you adjust the exposure. Try to get the meter to read +2 without blinking. Then increase the exposure by one stop. So for example if your shot is 1/50 s, f/10, ISO 3200, and reading +2.0 on the exposure meter, you would then change the aperture from f/10 to f/7, one stop faster, making the exposure +3.0. Then the exposure meter just blinks +2.0, but you'll know it's really +3.0.

But that presumes the exposure meter is accurate. Your extreme contrast scene with PP1 - 1/50, F4 reading +2.0 on the exposure meter, for example, is not correctly exposed despite the exposure meter reading. The only thing correctly exposed is the edge of the window. The indoor section is too dark and the outdoor section is too bright. If the indoor part is your subject, you need to expose for that.

I find the exposure meter to be quite accurate, but it's affected by how you aim the camera. Point the camera at the subject (the dark part of the room), and it read much lower than +2.0. It's reading +2.0 because you have the camera aimed at the bright edge of the window.

A more reliable (if less convenient) way to set the exposure is to use a standard picture profile (not S-Log2), judge the exposure on the LCD, and then change to S-Log2 and add 2-3 stops to the exposure. So for example, if you set a standard picture profile (not S-Log2) and use 1/50 s, f/4, and ISO 400 and you like the way the image looks on the LCD, changing to S-Log2 and ISO 3200 would be a 3-stop increase in exposure, which would be about right for minimizing the shadow noise while maintaining similar highlight range.

Yet another way is to use zebras and expose to the right on the histogram.

S-Log2 gives you more exposure latitude, but it doesn't remove the need to set the exposure correctly. It just gives you more room to be wrong with your exposure and still correct it in the workflow.
set wrote on 5/12/2015, 5:13 PM
A more reliable (if less convenient) way to set the exposure is to use a standard picture profile (not S-Log2), judge the exposure on the LCD, and then change to S-Log2 and add 2-3 stops to the exposure.
Even for lighting setup, this way is more preferred, and we can focus on doing grading in post rather than fixing it first.

Set
farss wrote on 5/12/2015, 6:38 PM
Just a few random thoughts:

An incident light meter might be a worthwhile addition to the kit bag.

Even those who shot film knew they had to wrangle lighting. Today we have cameras as good if not marginally better than film. That doesn't mean we don't need to deal with the same issues as before.

In very high contrast scenes I've found the quality of the lens can be a significant factor. I recently was forced to shoot some talking heads with extreme backlighting. There's considerable "bleed" around the edges of the subject. I've also found from first hand experience that one shouldn't overlook what a matte box with flags can do for the image.

Bob.
set wrote on 5/13/2015, 2:15 AM
Just in case anyone wants to play around with samples from High End Camera ARRI Alexa:
http://www.arri.com/camera/alexa/learn/alexa_sample_footage/

*Weird, I can't connect... is the FTP still available?
missylee wrote on 5/19/2015, 2:21 PM
How would you import Sony RAW files from an F55 into Sony Vegas. It is my understanding this can't be done presently. A Sony F55 rep. told me by phone this would be a feature available in Vegas this fall. Had you heard anything about this?
balazer wrote on 5/20/2015, 2:22 AM
Catalyst Browse and Sony Raw Viewer can both convert F55 raw footage to MXF XAVC Intra files in the S-Gamut/S-Log2 color space, which Vegas can read. I'll add S-Log3 to my config at some point also.

You're supposed to be able to export to OpenEXR files in the ACES color space, but it's not working correctly. I suspect that Vegas is performing an inappropriate color space conversion upon reading the OpenEXR files based on their metadata.
set wrote on 5/26/2015, 6:11 PM
Your blog site is mentioned by Sony Software in their Facebook page :)
https://www.facebook.com/SonySoftware/posts/10155655296915118
balazer wrote on 6/3/2015, 2:17 PM
I have posted new versions of my configurations with added color spaces:

* ACES RRT (Rec.709 video file full_range)
* Panasonic GH4

The Rec.709 full_range color space is the one to use if your camera records to a 0-255 luma range. Check Nick Hope's topic: Survey: What min/max levels does your cam shoot?

Big thanks to Nick Hope for providing the measurement and validation videos that I used to build the Panasonic GH4 IDT. See the GH4 IDT in action here:


The GH4 is showing good exposure latitude. The useable exposure range is from about two stops underexposed to 2/3 of a stop overexposed.


I also have a GoPro Protune native gamut matrix that I'd like to get tested. Please e-mail me if you have a GoPro HERO4 and you're willing to help.
videoITguy wrote on 6/3/2015, 5:04 PM
Looking forward to using ACES RRT fullrange. Thanks for your hardwork on this topic.
Satevis wrote on 6/3/2015, 5:22 PM
In a more typical 8-bit workflow, the Vegas Preview window is not WYSIWYG. We either have to a) rely on a secondary display and ensure that "Adjust levels from Studio RGB to Computer RGB" is checked in the Preview Device preferences, or b) Apply a temporary "Studio RGB to Computer RGB" FX preset and remember to remove it before rendering.

By the way, b) is exactly what my Preview Levels extension (http://www.semw-software.com/extensions/) does: Add a temporary effect that can be controlled from the preview window and will be bypassed automatically during rendering.
wwjd wrote on 6/3/2015, 11:01 PM
balazer, do any of these addins and replacement parts affects using Vegas normally at all? Can I keep doing my good old standard 8-bit/32-bit stuff like normal and only switch to ACES as desired after replacing all those files?
balazer wrote on 6/3/2015, 11:04 PM
When you turn off ACES by setting the view transform to "Off", the OpenColorIO configuration is not used and Vegas behaves normally. Media and rendering color space settings will have no effect.
balazer wrote on 7/13/2015, 1:31 AM
I have updated my configurations with new color spaces:

* Sony Cine1 (HyperGamma 4)
* Sony NEX-VG20
* Rec.709 (standard, without the ACES RRT)

The Sony Cine1 gamma curve has an extended highlight range, like S-Log2. But unlike S-Log2, Cine1 is designed to be similar to Rec.709, making it easier to grade, preview, and judge exposure with. Read Why Sony Cine1 (HyperGamma 4) is Better than S-Log.
ushere wrote on 7/13/2015, 3:00 AM
@balazer'

i have finally gotten around to watching your videos.

all i can say is MARVELOUS. i haven't watch such precise, informative and educational content delivered with such ease and clarity in many years. THANK YOU.