Thread: VisionColor introduces Spektrum

Page 3 of 4 FirstFirst 1234 LastLast
Results 21 to 30 of 31
  1. #21  
    Senior Member
    Join Date
    Sep 2013
    Location
    Long Island
    Posts
    1,451
    I also think it's worth mentioning that your exposure may greatly affect LUT use (in general) as well.
    Reply With Quote  
     

  2. #22  
    Senior Member DPStewart's Avatar
    Join Date
    Feb 2015
    Location
    The Desert
    Posts
    2,951
    cchriswake,

    I've never seen a man beat the snake before........ WHO ARE YOU?
    Cameras: Blackmagic Cinema Camera, Blackmagic Pocket Camera (x2), Panasonic GH2 (x2), Sony RX100 ii, Canon 6D, Canon T2i,
    Mics: Sennheiser, AKG, Shure, Sanken, Audio-Technica, Audix
    Lights: Every Chinese clone you can imagine
    Reply With Quote  
     

  3. #23  
    Quote Originally Posted by cchriswake View Post
    Let's talk about this.. and some other things regarding VC.

    First thing: Vision Color luts are not "paste and apply method" luts you find these days (i.e filmconvert, lightworks, etc). They are mathematically engineered, yes Engineered, to emulate the color space of Film.

    What does that mean? It means you can't just shoot any footage with any f-stop/luminance values and just slap a lut and expect great results.

    Vision Color explicitly states when you purchase any of their products, particularly the impulz version, you need to treat as if you're shooting on that particular film stock.

    Let's look at a sample workflow I've seen people on youtube/vimeo post when they use these luts wrongly:

    Example A) They typically set up their editor (resolve) and right away go to their footage on the color page.

    They do their primary grade and secondary's with the bmdfilm or log format they have. Then they "slap" a lut on a serial node. The footage looks quite crap. Why?
    Because they made all their corrections prior to the lut without taking into account that these luts have their own color space/gamma/values. This is one popular way that they're used incorrectly.

    Another is when they first put the lut in the node tree, which is fine. But they put the fc or fpe lut on the footage. Visually, it looks fine and the footage certainly pops out. But what happens when you start grading? The footage becomes noisier, looks compressed, looks like shit. Why?
    Because any grading should be done BEFORE the lut(s) are applied.

    But wait a minute.. didn't I contradict myself with the first example? A-ha, therein lies where the real culprit of issues stems from. You NEED to have an output and/or input lut already placed in your 'color management' tab in the setting manager BEFORE you start any grading process.

    Let's take a look at how a correct way for Impulz should go:

    Example B) Before we delve into the right way, let's look at what vision color offers in their package.

    This is what output/gamma they offer, in their own words -

    Film Contrast (FC): Non-linear distribution of image saturation with a film-like gamma curve.

    Film Print (FPE): Industry standard Vision 2383 film print. Medium contrast/saturation print stock.

    VisionSpaceTM (VS): Our custom gamma response which is ideal for starting your grade. Just enough contrast to get you started in the right direction without blowing out highlights or shadows and the same non-linear saturation as our film contrast profiles.

    Cineon® LOG (CIN): Industry standard Cineon LOG compound gamma (95-685). This output emulates the gamma response of the DPX scans from the film scanners (Hasselblad Flextight X5 & ArriScan 4k btw). Resulting negative emulations are compatible with all film print emulations that expect a true cineon curve as input and provide a perfect starting point for fully fledged color grading.

    **Important Note** = you can only use ONE of these for your color space. I have read people who cross an FC lut with an FPE lut on the same footage and wonder what the hell happened to their image(s). If you use two (you should never use two or more) luts like these in the same image, you have to stretch the image beyond its color space to make it even look normal, which by that point you'll have destroyed the image anyway, even if you shot on raw.

    FC, FPE and VS you can put as a singular output in Color management settings under the 3d output. However, I found the best AND (this is important) accurate way to use these luts is for their cineon values. No company out there has come close to what VC has done emulating the film DI process when it comes to Cineon. I'm going to skip explaining what the Cineon process actually is regarding the DI due to time constraints. I will say this about the process: you can't expect to shoot and expose how you would normally for your digital camera whether it's log, raw, slog, etc. You need to take into consideration that the "emulation" in their marketing isn't geared for people to use it in any setting, it's for people who want to shoot on film like stocks and have the same process when it's time for color grading.

    Take a look at some of the stocks they have, particularly the current industry used Kodak negative stocks such as -

    - Kodak Vision3 50D 5203, Kodak Vision3 250D 5207, Kodak Vision3 200T 5213, Kodak Vision3 500T 5219, etc

    Take a look at this page now and read through the different manuals for these same negative stocks: http://motion.kodak.com/motion/Produ...on/default.htm

    Did you see all the things you have to take into consideration before shooting on those stocks? How you need to think about the T-stop, brightness-luminance values, indoor/outdoor qualities, etc for these stocks?

    I don't know what your footage looked like. But I'm betting you didn't really think about lighting your frames with the exact stock in mind. For example, for a 500T 5219 Kodak negative stock you need to shoot in an indoor setting/studio with tungsten lighting (hence the 'T' in the name) or even a daylight interior setting with window light as your main source for filming. This is NOT a good stock to shoot on outdoors where light leaks, wild lighting can occur.

    The noise that you encountered is the same thing as when you stretch an image a bit beyond its limit in the light values. So if you shot your footage with a setting or T-stop that looked good on raw on your monitor, and you got significant noise in parts of your image after putting say, a 5219 lut is because the negative stock didn't match with its intended setting. This is where a good monitor where you could preview luts comes in, such as an Atmos, smallhd or even the BM assist.


    Back to cineon. It's the best way to use the luts due to the results it delivers in terms of emulating the film gamma.

    Impulz comes in 16 negative stock options, including the aforementioned Kodak stocks.
    It then comes with 7 Film Print options for delivering.

    Simplified version -> You can put any of the 16 negative stock in the 3d lut Input section under project settings.
    Then you will put any of the 7 Film Print options in the 3d lut Output section under project settings.

    BUT, once again. You must know what stock you'll want to shoot and learn its values. Again I'll mention, when they said they engineered the luts to follow their real life counterparts, they weren't joking. It's work, but that's what shooting on film is for DPs and colorists. 3D Luts are not meant to be an easy, quick application that solves all image problems. When you pick a stock you'll know you want to emulate, and then expose shots/light accordingly, you'll be astonished the world of difference it can do.


    Hope this helps some. Sorry if it came off a bit condescending but it drives me mad that with all the info out there, there's this misconception that 3D luts are magical paintbrushes that makes any footage look like film and perfect.

    *Also, regarding Spektrum my friend says it should be out for beta around end of next month and released for January 2017. Though it could possibly skip the beta and go straight into release a bit earlier. All they need to do is refine and test for all digital cameras that are active from the Alexa all the way to the Canon Cinestyle profile.
    One of the best posts on here. Bravo.

    Naturally, a few questions if you don't mind..

    I've taken your advice and have begun scouring any sources I can find in relation to exposing/shooting for specific stocks of film. Right now, I am focusing on VISION3 500T Color Negative Film 5219/7219.

    From what I've gathered:
    Film stocks are rated at a certain ASA and on top of that have their own Exposure Index (EI). The higher a film's rating, the more sensitive it is to light. The EI doesn't have anything to do with changing the ISO/ASA on the camera, but rather your light meter. As far as I understand right now, this is what "treating film stock" on set when you're a cinematographer - using a method known as "pushing or pulling" the film (Not to be confused with push processing, which has to do with the actual development of the film)

    5219 is rated EI 500 for tungsten, and EI 320 for Daylight.

    So for the sake of simplicity if I'm shooting a scene on 5213, and I set my lightmeter for 1000 ASA, I am underexposing by one stop since I am treating the film negative as if it was a 1000 speed film. Meaning the ASA on my BMPCC does not change. If I set my lightmeter for 250 ASA, this means I am overexposing by 1 stop.

    If I'm right so far, my confusion right now lies in how you "rate" a film stock, specifically 5213 Cineon. I've read that generally you can't underexpose more than a stop for film, but when OVERexposing a stop it gives you about 2 stops of latitude either way.

    So in theory (and stop me as soon as I'm wrong), shooting raw on BMPCC, I set my ISO to it's native 800. I light my talent with tungsten fresnels, indoors. If 5213 is rated at EI 500, i set my light meter to ASA 500, shutter speed to 180, and a reflected reading off the talent (being lit by the key) and out comes my exposure. Now let's I wanted more detail and less noise in the shadows, I rate my 5213 film stock at EI 250, take another reading, and out comes the new aperture - effectively pushing the film and overexposing by 1 stop. Here is where I'm getting confused - if I am already overexposing for the film stock by a stop, where does ETTR come in regard to making the most out of the BMPCC's dynamic range? And how is EI determined for specific film stocks?

    Sorry if this is very messy, but there has been tons of new information coming in and my brain is slowly turning to mush. Any help is appreciated!

    Edit: Just came by this very excellent post on the forums of Cinematography.com-

    Quote Originally Posted by David Mullen ASC
    The recommended ASA by the manufacturer means that you will end up with a negative, if developed normally, of average density with the widest range from bright to dark information that the negative can hold.

    As you overexpose a negative, you expose more of the smaller (thus less sensitive) grains as well as the larger ones that got exposed originally. The more you underexpose, only the larger (thus more sensitive) grains have time to get exposed.

    So some slight overexposure leads to some tightening up of the grain structure, which makes the image look less grainy (though the large grains are always there.) It also exposes more shadow information at the expense of some bright highlight information (no free lunch.) When this slightly overexposed negative is developed normally, then printed down, the blacks tend to look richer in the print, and thus the contrast looks snappier, which in turn makes the image look a little sharper, plus the colors look richer because of the better blacks, etc. Now if you're not printing, there is less benefit to overexposing.


    By the way, rating a stock 1/3 of a stop slower is so conservative that it's within a margin of error, i.e. you could easily accidently underexpose a shot by 1/3 of a stop and counteract the slower rating. So if you want to rate a 500 ASA stock at 400 ASA, you could do so without too much worry.
    Last edited by iwander; 12-05-2016 at 10:58 AM.
    Reply With Quote  
     

  4. #24  
    Senior Member
    Join Date
    Apr 2012
    Posts
    219
    You are overthinking the process.
    A few notes:
    1) You don't use the LUTs to get a technically correct translation of the negative + print process. You use them as a creative tool. You don't need to expose your footage "correctly", and besides, there is no such thing as correct negative exposure. Negative is (was) universally overexposed anyway, so that there is enough leeyway under. You'd then bring it down in processing + printing by adjusting the printer lights accordingly.
    2) You CAN'T get a technically correct translation anyway. Even a sensible approximation would need a prohibitively large LUT size.
    3) You will get noise (and banding) from film LUT application for a variety of reasons: not enough precision in the source image, especially in the darks; not enough LUT precision (LUT size); highly non-linear behavior in film darks and whites which requires LUT smoothing to minimize banding artifacts (practically all technically calculated film LUTs will need smoothing).

    What can you do then?
    1) Don't use low precision footage.
    2) Expose sensibly. Underexposure should be banned from your life. Overexposure is almost always a good idea cause you get a denser image = less chance for banding/LUT noise later on. Most log curves have a known distribution and can be linearized, exposure compensated while linear and then log-ed back for film LUT application (preferably using a function, not a LUT, i.e. dctl or a similar tool). Alternatively, do exposure compensations using the offset and log tools.
    3) Use tetrahedral LUT interpolation (there is a setting in Resolve for this): for log/gamma encoded RGB footage tetrahedral interpolation will usually have less distortions/artifacts than the default trilinear interpolation.
    4) Do the majority of your grading under the print lut using log controls.
    Reply With Quote  
     

  5. #25  
    Quote Originally Posted by cpc View Post
    You are overthinking the process.
    A few notes:
    1) You don't use the LUTs to get a technically correct translation of the negative + print process. You use them as a creative tool. You don't need to expose your footage "correctly", and besides, there is no such thing as correct negative exposure. Negative is (was) universally overexposed anyway, so that there is enough leeyway under. You'd then bring it down in processing + printing by adjusting the printer lights accordingly.
    2) You CAN'T get a technically correct translation anyway. Even a sensible approximation would need a prohibitively large LUT size.
    3) You will get noise (and banding) from film LUT application for a variety of reasons: not enough precision in the source image, especially in the darks; not enough LUT precision (LUT size); highly non-linear behavior in film darks and whites which requires LUT smoothing to minimize banding artifacts (practically all technically calculated film LUTs will need smoothing).

    What can you do then?
    1) Don't use low precision footage.
    2) Expose sensibly. Underexposure should be banned from your life. Overexposure is almost always a good idea cause you get a denser image = less chance for banding/LUT noise later on. Most log curves have a known distribution and can be linearized, exposure compensated while linear and then log-ed back for film LUT application (preferably using a function, not a LUT, i.e. dctl or a similar tool). Alternatively, do exposure compensations using the offset and log tools.
    3) Use tetrahedral LUT interpolation (there is a setting in Resolve for this): for log/gamma encoded RGB footage tetrahedral interpolation will usually have less distortions/artifacts than the default trilinear interpolation.
    4) Do the majority of your grading under the print lut using log controls.
    I guess I took his advice too literally. Although, after going down the rabbit hole of film stock was pretty fun. Definitely have some tests to do tomorrow morning regarding the IMPULZ process.
    Reply With Quote  
     

  6. #26  
    Quote Originally Posted by iwander View Post
    I guess I took his advice too literally. Although, after going down the rabbit hole of film stock was pretty fun. Definitely have some tests to do tomorrow morning regarding the IMPULZ process.
    What are you both referring too? I searched a couple documents but none really told me how to expose different stocks. Am I missing sth?
    Reply With Quote  
     

  7. #27  
    Quote Originally Posted by Rotbart View Post
    What are you both referring too? I searched a couple documents but none really told me how to expose different stocks. Am I missing sth?
    cchriswake provided a link to Kodak's main page where you can read about each film stock individually.
    Reply With Quote  
     

  8. #28  
    Reply With Quote  
     

  9. #29  
    Senior Member LochnessDigital's Avatar
    Join Date
    Jul 2014
    Location
    Tucson, AZ
    Posts
    431
    Aaron Lochert
    Reply With Quote  
     

  10. #30  
    Senior Member Frank Glencairn's Avatar
    Join Date
    Apr 2012
    Location
    Germany
    Posts
    3,686
    cricket.... cricket... tumbleweed cricket...
    Reply With Quote  
     

Similar Threads

  1. Highlight Recovery Introduces Magenta Noise Pattern
    By APPROD in forum DaVinci Resolve
    Replies: 10
    Last Post: 06-05-2015, 04:35 PM
  2. VisionColor Osiris LUT
    By AraM182 in forum General Discussion
    Replies: 9
    Last Post: 04-01-2014, 03:49 PM
  3. VisionColor OSIRIS LUTs
    By Unknown in forum DaVinci Resolve
    Replies: 27
    Last Post: 03-30-2014, 02:08 PM
  4. Visioncolor M31 Cinema LUT
    By Unknown in forum Workflow
    Replies: 8
    Last Post: 07-15-2013, 10:45 PM
  5. Replies: 82
    Last Post: 04-05-2013, 02:15 AM
Bookmarks
Bookmarks
Posting Permissions
  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •