resolve workshop details 2016 - concordia university · pdf filedavinci’resolve’11...
TRANSCRIPT
DaVinci Resolve 11
(Page numbers without a name reference in this document refer to the Resolve 11 manual, found in the Applications/ DaVinci Resolve folder)
This workshop covers some basic features of Resolve. It really just scratches the surface of this very detailed application.
Resolve is primarily a color grading (correction) software, but it is also a useful basic video editor and can accept all RAW video formats.
At the CDA we use Resolve for the following tasks:
-‐‑color grading a sequence sent from Premiere and then either exporting this graded sequence as one Pro Res or DPX file, or as a series of Pro Res files for relinking in Premiere
-‐‑transcoding RAW video into an offline video format (typically Pro Res) for editing in Premiere – online RAW video is then linked again in Resolve for color grading
There are two versions of Resolve:
Resolve Lite: the free version. This does not include noise reduction tools and has a timeline resolution cap of 3840 x 2160. Resolve Lite is installed in all the CDA edit rooms.
Resolve (Full version). This has noise reduction tools and no resolution cap on the timeline. It’s expensive but it is packaged with Black Magic Cinema Cameras (but not the Pocket camera). The CDA has one copy of the full version in edit room zero.
You may go back and forth between the two versions as long as you comply with these restrictions (you can’t open a 4K timeline with noise reduction in the Lite version for example).
Resolve works on almost any computer, but the performance may vary according to your source media.
Resolve 11 System Requirements:
Mac OS X 10.8.5 or later (or Windows)
12 GB RAM recommended, 8 GB minimum
Other Requirements:
A calibrated screen
A second monitor really helps (for your video scopes)
A good graphics card
Expanded Systems:
Open FX Plug-‐‑ins are available (for example: film grain effects)
A second video card dedicated for rendering and effects processing: Resolve works with CUDA (NVIDIA cards) and Open CL (AMD cards)
A Black Magic Design video card or thunderbolt device with HD SDI or HDMI output to a color correction monitor
A color correction monitor
A control surface
Read the system requirements carefully for expanded systems.
Part One: Transferring an Edit from Premiere to Resolve We can edit in Resolve but typically the editing happens in another software, such as, Premiere.
XML export from Premiere
You can export a sequence from Premiere to a timeline in Resolve using XML export.
Other project file exchange types are also compatible with Resolve: Avid AAF files or generic EDL files.
Sending a Sequence from Premiere to Resolve using xml:
-‐‑if your Premiere sequence is set to the NTSC frame rates of 23.976 fps or 29.97 fps then the sequence time code should correspond to your source media: if the source is drop frame time code then the sequence should also use drop frame time code (you can change this in the Sequence Settings in Premiere after the sequence is created)
–in drop frame time code, the 00 and 01 frame are dropped every minute except for every 10th minute
-‐‑duplicate the sequence and clearly name it as the sequence to be corrected – when you import the xml file into Resolve you will be asked to choose the sequence to import – so it helps to clearly identify it
-‐‑delete blank video tracks and audio tracks
-‐‑VERY IMPORTANT: remove color correction effects on all clips otherwise Resolve will translate this as an effect
-‐‑keep the video on video track one, unless compositing layers
-‐‑move generated media (Premiere bars, black, countdown leaders and text generated in Premiere) and unsupported media (AE Linked Comps) to an upper track and turn OFF the track
About Effects
-‐‑generally you want to send the media to Resolve without effects (but there might be some exceptions)
-‐‑there is a list of supported effects on page 368 of the Resolve manual when using XML import
-‐‑transitions (dissolves)/opacity/composite modes/position/scale/speed effects are supported
-‐‑Resolve claims to accept mixed frame rates in a sequence but in my experience this causes problems with xml import – so avoid mixed frame rates
-‐‑mixed frame sizes are accepted (but you can rescale in Resolve if the scaling is not correctly interpreted)
-‐‑keyframed positioning, scale and opacity is supported if the values are typed in manually in the effects window (i.e.: scale to frame size does not work)
-‐‑linked audio or merged clips are supported
-‐‑nested sequences are not supported
-‐‑some still formats are not supported
-‐‑ speed changes are supposedly accepted but in fact they cause problems
-‐‑Resolve has a speed tool that is better than Premiere but if you want variable speed effects, I recommend “baking in” variable speed effects using the time warp effect in After Effects
-‐‑unsupported effects can be handled in two ways (think about whether the effect should be applied before or after color grading):
a) the effect can be “baked” into the file (by exporting as Pro Res HQ) and then reimported into Premiere to replace the original clip before the xml export
b) the effect can be removed and then applied later to the corrected clip that is exported from Resolve back into Premiere –a timeline can be exported as individual clips from Resolve for relinking in Premiere
Exporting a Reference Movie From Premiere
-‐‑export a reference movie to verify the xml export worked correctly
-‐‑a reference movie can also contain color correction done in Premiere as a suggestion to the “colorist”
-‐‑if you wish, apply the timecode effect to the entire sequence to have a time code burn-‐‑in, otherwise a visual reference is just as good
-‐‑ export the reference movie as Pro Res Proxy or something similar
Part Two: Setting Up a Resolve Project These steps are covered in a series of video tutorials:
http://cda.concordia.ca/cvvt/CVVT/Resolve_10/Resolve_10.html
Resolve needs a database folder where it will keep your user and projects.
1. Before you open Resolve, create a folder on your external drive. For convenience call this folder something simple like EXTDB, or ResolveDataBase. You will have to recreate this database each time you switch edit rooms in the CDA, so the name of the folder should be the name of your database.
2. The database is created with the stack of pancakes icon in the first window of Resolve. You are creating a DISK database. Make sure this database is selected before you go back to the User creation window.
3. Then create a User. Resolve will not prompt you if you lose your password. So pick a password that you will not forget.
4. Create a new Project.
5. Once the project is open, go to (Top Menu) DaVinci/Preferences/Media Storage. Create a new Media Storage location that is your database folder on your external drive.
6. Then go to the Project Settings (the wheel icon in the bottom left of the interface). In Project Settings/General Options/Working Folders, set the Cache File and the Gallery Stills location to the same Database folder. The folders are hidden but they exist in this location. This setting is important so you do not lose preview files and gallery stills when you move to other computers.
7. Save Project. Quit and Reopen Resolve to activate your preference changes.
Steps two and then steps five to seven have to repeated each time you work on a new computer. You cannot see your user and projects until the database has been recreated for that version of Resolve.
Note that Resolve will not see any folders in the Finder that you create once Resolve has already been launched. You can always see to the root level of your drives with Resolve, so make sure you are not placing folders in a location for which you lack permissions.
Importing the Reference Movie and XML file into Resolve
Create a New Project in Resolve.
Check the Project Settings/Master Project Settings/Timeline Format. You must match the image size and frame rate of your Premiere project. Change the size and frame rate in the Resolve project settings, and then make sure to save the project to initiate these changes.
Import the xml file: if you have multiple sequences in the project, make sure you are importing the right sequence. By default your media will also be imported into the media pool.
Import the reference movie into the Media Pool by right clicking on the clip in the library and importing as “an offline clip”. It has a checkered icon.
Check sync with the reference movie. Select your timeline icon in the timelines window in the top left of the edit page and right click “link offline edit clip”. Then go to the media pool window and open the reference movie in the viewer. In the viewer, bottom left, select the checkered icon. The viewer and timeline window are now ganged for checking sync.
If the sync is correct, go to the Color workspace to start grading.
Part Three: the Resolve Interface Resolve has four workspaces: media, edit, color, and deliver.
1. Media Workspace
Library Window: allows you to navigate into your hard drives to import clips
-‐‑the globe icon indicates the database volume on top of the Library window –this location is determined in DaVinci/Preferences/Media Storage
Note that you cannot see a folder on any drive that was created after Resolve launched
Metadata window (bottom right): click on a clip to see the resolution/frame rate/codec etc.
Media Pool: create bins to contain media – this is your storage area, like the Project window in Premiere
Some Import Options:
Scene Cut Detection –really useful if all xml/aaf/edl efforts are failing
-‐‑on import you can right click on a clip for “scene cut detection” –this allows an edited video to be segmented into cuts to be individually graded
-‐‑no new media is created
-‐‑this is a last ditch resort to import a sequence into Resolve –use prudently and with caution-‐‑ of course it does not work with cross dissolves-‐‑ straight cuts only
Importing Image Sequences:
Image sequences come in at the frame rate specified in Project Timeline settings but the frame rate can be changed after the import by right clicking on the file and going to clip attributes.
Video Clip frame rates can be changed in the same manner.
2. Edit Workspace
A project can have multiple timelines at different frame rates and resolutions.
Resolve has a simple but effective editing workspace.
All familiar edit controls are present:
Insert and overwrite edit buttons
Tracks and track targeting related to the above buttons
Timeline view options
Trim Edit Tool for Ripple/Roll and Slip Edits
Toolbox of Transitions/Titles/Generated Effects
Audio Meters
Speed Changes and Retiming Changes (variable speed changes) are possible and look very good.
-‐‑ but video playback is not as smooth in Resolve as in Premiere
Two Options if your system is slow (if playback is not smooth):
Option 1. Background Render Caching
– a cached frame is like a preview file in Premiere (or a Render file in Final Cut)
-‐‑ set the cache frame location when you create a project (Project Settings/General Options/Working Folders)
-‐‑in the same window in General Settings you can also choose the format for Cached Frames: Pro Res or DNxHD are good choices
-‐‑the cache files will be invisible but they will still take up space on your hard drive
-‐‑on the top menu choose-‐‑ Playback/Render Cache/Smart – portions of the timeline that play back poorly will be cached automatically
-‐‑the red line indicates portions of a timeline to be cached
-‐‑Blue status bars indicate when a clip has been cached on a timeline or color page
-‐‑you can also cache a clip manually by right clicking and selecting Render Cache Clip Source/On
-‐‑the cache folder can be deleted at any time from the top menu Playback/Delete Render Cache
Option 2. Making Proxies of DPX media
-‐‑lower resolution video files of DPX media are called proxy files
-‐‑Project Settings/Image Scaling determines the resolution of Proxies.
-‐‑this is also where you enable proxies.
-‐‑the location of the Proxy files is determined by Project Settings/General Options/Working Folders. Set the Proxy file location to a folder on your hard drive. They will be hidden files in the Finder.
Resolve will generate proxies in the background or you can also generate your own proxies with the Proxy Manager (right click on a clip in the media pool).
I assume that the proxy file format is a lower resolution DPX file.
3. Color Workspace
Obviously this is the main part of the application.
Without going into detail on the theory of color grading, here some of the basic tools in the Color workspace:
Scopes
Top Menu: View/Video Scopes
You want to display the scopes while making corrections. It helps to have a second monitor since they appear in a “floating” window.
In the scopes window you can display one to four scopes at once. Make sure to adjust the brightness controls (bottom right) on the vectorscope, otherwise the pixels are not visible. Each scope has separate controls.
Monitoring the image on a computer monitor:
Toggle Viewing controls:
Option F: Enhanced Viewer
Shift F: Full Screen
Adding a Node
In resolve corrections are applied to nodes: a tree like structure can develop with each node influencing the successive node. You can think of a node as a layer.
Get out of enhanced viewing mode. The Node window is on the top right.
Keep the first node blank. Add a node as you add changes.
Top Menu: Nodes/Add Serial Node
The first node is typically an overall primary color adjustment.
You can try out different color adjustments using different nodes and turning off selected nodes.
Command D: turns off a selected node
Option D: turns off all nodes
Parallel Nodes can be added to combine different effects simultaneously (for example three mattes simultaneously applied).
There are other types of nodes for specific functions.
Left Palette: Color Wheels
Most of the action on the color workspace takes place on the color wheels.
The color wheels have two settings (top right): Primaries and Log.
The Primary option is used for correcting linear (HD) video, the Log option is for correcting RAW video or log sampled video.
Primary Color Control Wheels (P.517)
Auto Contrast Button (bottom left): try it, it might work.
Lift/Gamma/Gain/Offset
If you have used the three-‐‑way color corrector in Premiere then these controls will look very familiar. These controls correspond to Shadows, Midtones and Highlight adjustments in the three-‐‑way tool. Lift and Gain are linear adjustments, while gamma is nonlinear. There is overlap between the three areas of luminance.
The wheels on the bottom are for contrast adjustments.
Start with the lift and find your black point, then adjust gain and gamma.
Going to the left on the Lift wheel lowers the black point. Going to the right on the Gain wheel raises the white point.
The color areas are for hue adjustments.
The Offset wheel makes a general correction to entire hue or luminance.
Below the wheels you will see other controls:
Contrast and Pivot
These controls are an alternative method for controlling contrast adjustments. The pivot controls the center point for contrast adjustment and the contrast stretches out the pixels from that point over the entire area of luminance.
Saturation/Hue/Luma Mix
These are global controls for the entire image.
Luma Mix: At 100: automatic adjustments are made to maintain luma levels when doing individual color channel adjustments. At 0: luma levels will change with individual color channel adjustments. (p.522)
Comparing Video Clips
Grabbing Stills
Stills are used for comparison between shots or alternative grades.
Right click on the image in the canvas window and “Grab Still”. The stills gallery is located to the left of the canvas window. You can rename the still. Make sure that you have set your still location in General Options of the Project Settings.
To see the still next to the image to be corrected: select the still in the gallery and then right click on the video you are correcting in the canvas window and choose “Show Reference Wipe”.
Copying a Grade to Another Clip
There are many ways to copy grades. Here I will mention only a few.
To copy all the nodes from one clip to another:
Grab a still of the finished grade from clip A, then when your new video clip B is selected, right click the still A from the gallery and choose “append node graph”.
Copying individual nodes:
Copy the node from one grade and paste it to a new node in another clip.
Using Memories:
Select a clip from the timeline and from the top menu Color/Memories/Save Memory (A to H).
Select another clip and Color/Memories/Load Memory (A to H)
There is a memories gallery in the top left of the Color workspace.
Right Palette: Curves (P.552)
For the Photoshop users these might be more familiar. You can get more specific adjustments from the curves.
By default the curves are ganged together, this can be turned off from the menu in the top right of the curves window.
There is a zoom button in the middle of the curves window.
You can click on a pixel in the canvas window with the eyedropper and automatically this value will appear as a point on your curves.
Initially you will see the Custom curves but there are six other curve options:
Soft clip, hue vs hue, hue vs sat, hue vs lum, lum vs sat, sat vs sat.
The soft clip window can be used to avoid unpleasant clipping in highlights or shadows. You can also create soft clip LUTS in Resolve to apply to all clips.
The Hue vs Sat/Hue/Lum curves allow you to isolate a particular hue in the image to apply a correction. This is a secondary correction and works a bit like a matte but in each case you are adjusting hue, saturation or luminance of that particular hue.
The Sat vs Sat curves can isolate an area of the image based on saturation of hue.
The Lum vs Sat curves can isolate an area of the image based on luminance.
In all cases use the eyedropper to identify the source and then refine the curve.
Secondary Correction in the Color Workspace
Qualifier (P.587)
Secondary color correction refers to correcting only one part of the image. The qualifier is a traditional secondary color correction tool. A matte is created with the eyedropper. One color in a particular area of saturation and luminance can be used to create a matte.
Use as follows: select the object in your image with the eyedropper, turn on the “magic wand” icon to see the matte. Select the plus and minus eyedroppers to continue refining the matte. Change the values in the hue, sat, lum scales to refine as well.
Any subsequent change in the primary color wheel window or curves will pertain only to this matte.
Shape Window
These are like After Effects masks.
There are preset shapes to which you cannot add points and there are polygonal or curved shapes that can be drawn by hand. The polygonal shape is the most useful.
The softness controls overall feathering.
If you want to reverse the shape (make changes outside rather than inside the shape) go to the key window (key icon) and click on the output toggle button (top right).
Tracking
Tracking can be applied to a matte shape.
Apply the shape to an object and then track backwards or forwards with the tracking window. Keyframes will be created in the window to the right.
The shape can also be manually manipulated in the window after the initial keyframes are in place.
It is best to track a small portion of your object and then increase the size and feathering of the matte as required during the shot. The sizing and the feathering will also be keyframed.
Keying
The keying window allows you to see the matte and make more adjustments.
It also allows you to reverse the matte.
Resizing Window
This is used for resizing images. Resizing can also be keyframed. The keyframe window is to the right of the right palette.
Keyframing Color Changes
Keyframing functions much like in After Effects. Turn on key frames for the parameters you wish to keyframe and then make changes in time. Each node will appear in the keyframe window as “Corrector” with a number attached.
Grading RAW and Log files in the Color Workspace
This is a bit of a perfunctory description of an extensive topic.
About RAW files and Log sampling:
All RAW video formats use some variation of the Log color encoding. This refers to “media with Log-‐‑C or similar gamma and color encoding, derived from the Cineon Log gamma curve, developed by Kodak to digitally store flat contrast, wide gamut image data that preserves image detail with a wide latitude for adjustment”. (P. 524)
Log encoding is used in “digital negative” video files or image sequences to approximate the dynamic luminance range of film. Linear sampled 8 bit or 10 bit YUV video ignores much of the brightness information in the world. Log sampled images preserve more highlights. Kodak developed logarithmic digitizing of images for film to digital transfers.
“With linear digitizing the brightness difference was constant everywhere and the relative brightness changed from sample point to sample point, getting larger towards the darks. With log digitizing the brightness difference changes and the relative brightness stays constant. From dark to light, one code value change represents a uniform change of brightness to the eye. Linear is uniform in brightness, log is uniform in perception.” (Wright, p.402)
Log encoding works to translate a greater dynamic range of luminance because our eyes perceive fewer changes in highlights than shadows. The shadow and midtone portion of the image is rendered in more detail than the highlights.
If you are working with linear HD video then you can start color grading immediately but if you are working with RAW video or log encoded video, you must first decide how to normalize these clips. You may have log encoded RAW files (RAW video is always log encoded) or log encoded files in a standard video codec like Pro Res 422 or some type of MPEG file used by Sony, Canon or Panasonic.
If you have RAW files you can apply a LUT at the project level. All RAW clips will automatically be normalized on input into the Media window. Clips without LUTs applied will look desaturated and washed out. Normalizing means that the clip has a contrast adjustment applied to interpret the clip into a certain color space or gamut (usually HD video Rec.709).
About LUTs:
LUT stands for look up table. This is a mathematical process. “A table of numbers that uses the value of the input data as an index into the table to look for a second number for output.” (Wright, P.460) The types of image processing done by LUTs in Resolve may vary but they are often used for log to linear conversions.
Gamut is a range of colors that is expressed in a technical name (Rec.709 for HD video).
Gamma refers to luma reproduction. It defines the black point. The current standards are:
2.6 – DCP projection
2.35-‐‑ EBU standard (European Broadcasting Union)
2.2 -‐‑ Consumer standard (computer monitors)
HD video color correction monitors use the 2.35 EBU standard.
Here are three options for interpreting RAW or log encoded files:
Option One: Managing RAW media at the Project Level:
Project Settings/Camera Raw:
On the top left of this window you determine how each type of RAW file (ARRI, Red, Sony, Cinema DNG, Phantom) will be interpreted. These settings affect all the RAW clips of that type in all the Resolve workspaces. For Black Magic Cameras, that shoot log images in Cinema DNG files or Pro Res, only the Cinema DNG files are affected by the Resolve Project Settings. Pro Res files are not RAW files.
For example, with Cinema DNG files her are the options:
Choosing to Decode Using Project/ Color Space Rec.709/Gamma Rec.709 will apply a LUT to normalize the clip to the HD color space. You may want to do this if you are generating low resolution clips for editing in Premiere or AVID.
Choosing to Decode Using Project/ Color Space BMD Film/ Gamma BMD Film is like having no LUT applied. You will be grading from the low contrast log image.
This is a good default starting position for grading since LUTs can also be applied to Nodes in the color window.
Choosing to Decode Using Project/ Color Space P3 D60/ Gamma P3 D60 is a digital cinema LUT. You could use this to see if your clips look correct when displayed through a Digital Cinema projector. The CDA does not have one of these projectors.
Choosing to Decode Using Camera Metadata will use the LUT supplied by the camera (this will probably be Rec.709)
Option Two: the RAW settings in the Color Workspace
For each individual shot, you can change the RAW settings in the Primary Color Correction Panel (Camera Icon). There are presets by camera type. Once again, if you shot with a Black Magic Design camera, this option is only available for Cinema DNG clips, not Pro Res clips.
Option Three: Applying a LUT on a Node in the Color Workspace
A 3D LUT can be applied to a Serial Node to any given clip. There are a number of 3D LUT presets for log to linear conversions for Arri, Sony and Black Magic camera video. You can also import custom LUTs into Resolve from other camera manufacturers, Canon for example.
I think this is the most useful way to apply an LUT. This method of applying LUTs can work with log encoded video files, like Pro Res or MPEG files.
Option Four: Grade from Scratch (without applying a LUT)
Use the Log Color Wheels for contrast adjustment and primary color adjustment.
It can be helpful to use a gallery still of a LUT preset as a reference when grading from scratch.
Log Color Control Wheels (P.523)
The color wheels in the left palette can be switched from Primary to Log mode for correcting log encoded video, RAW video.
The controls look similar but they respond differently and the names are now:
Shadow/Midtone/Highlight/Offset
Adjustments made with the Log color control wheels are generally made to the node before the one with the LUT applied, if a LUT is applied.
When correcting Log clips, start with the Offset wheel. Define the overall contrast using the slider to find the black point and then adjust the contrast and pivot controls to increase contrast. Once the contrast is roughly set, then adjust the overall hue if necessary.
Then go into the individual luminance wheels to affect shadows, mids and highlights. These adjustments can be minor and detailed.
Control the color cast for each of these areas.
Using a LUT as a reference
One trick is to first apply a default LUT to the clip and to grab a still. Then delete the LUT on the node but use the reference still as a guideline for your adjustments. The default LUTs can be too aggressive but they are useful as a comparison and to make sure to eliminate the low contrast haze from RAW log images.
Low Range/High Range
Adjusts the borders between the midtone adjustment and the shadows and highlights.
Using the FSI Monitor in Room Zero
The FSI monitor is the professional color grading monitor.
The FSI Monitor can represent different color spaces but it is essentially a YUV 4:2:2 monitor working in the Rec.709 gamut. This is the color space of HD video.
The monitor can simulate the P3 color space of DCP Projection but it is only a simulation, and therefore I leave this option off. When color grading for DCP, post-‐‑production studios will use a Digital Projector calibrated for the P3 color space. But the FSI monitor has a slightly cooler color cast, and therefore it will create a reasonably accurate output for DCP in addition to HD video.
You will notice that the main computer monitor in edit room zero has a slightly warmer color cast than the FSI monitor. Compare the two. The computer monitor will show what an output on the web will look like, or perhaps an output for LCD television.
Because the FSI monitor is set to receive a broadcast YUV 4:2:2 Rec.709 signal, it is important that your Video Output settings are correct in the Resolve Project Settings.
Project Settings/Master Project Settings/Video Monitoring/Video Levels should be checkmarked.
Do not checkmark 4:4:4 SDI output. Although the video card can do 4:4:4 output over 3Gbit SDI, the monitor is not set up to handle this input.
Display the waveform on the FSI monitor (Button F1) and compare with the waveform in Resolve when you are in Color Workspace (View/Videoscopes). Even
though one is in IRE units (FSI) and the other in 10 bit values (Resolve), they should correspond. This is what is important. The RGB values will be remapped correctly.
In Resolve, the waveform monitor shows luminance values from 0 to 1023. This is the total luminance value range for 10 bit video.
Broadcast YUV video has a decreased range of 64 to 940. This is what the FSI monitor is representing. Despite the difference in scale, you should see the same pattern and your black levels should line up, as should your highlights at 100 IRE on the FSI or 1023 on the Resolve monitor.
In the Resolve media pool you can look at clip attributes for each clip. Leave the level for each clip on AUTO. Do not specify video or data levels.
Also, in the Video Monitoring/Video Format selection, do not use progressive segmented frame video formats on output. If your video is 1080 24p, use the 1080 23.976fps setting not the 1080PsF 23.976. This will stabilize the signal when paused.
4. Deliver Workspace
This is the export workspace.
The most common export options are:
-‐‑ a single Pro Res HQ file of the entire timeline
-‐‑multiple Pro Res HQ files for relinking to a sequence in Premiere
-‐‑a DPX file of the entire timeline for creating a DCP.
The Deliver workspace can also be used for creating lower resolution files (“offline” files) of RAW media for editing. Pro Res Proxy is a good format for offline files. When creating offline versions or multiple Pro Res files for relinking, make sure to Use Source Filename in the File/Save as options. Offline and online versions must have identical file names (but different file extensions).
Multiple jobs can be added to a render queue.
There are Easy Set-‐‑ups for Pro Res (Final Cut Pro Round Trip) and DNxHD (AVID Round Trip) and H.264 (Video Sharing) but the Basic individual render settings are fairly self-‐‑explanatory. The Intermediate and Advanced windows add more settings, like manual video/data level adjustment.
Resolve Deliver settings:
If you are working on a Mac, the two main export formats from Resolve are ProRes HQ for a video master (back to Premiere or Final Cut) and a DPX export for creating a DCP. In addition, the MXF DNxHD export option should be used for a video master going back to AVID. I assume that what I say below about a ProRes HQ export also pertains to a DNxHD (but I have not actually verified this empirically).
To maintain the correct luminance levels on export you must be aware of some of the differences between HD video and RAW video. There are two major differences:
The first difference is color compression.
Although we speak of video as using the RGB color system, with three channels of information (Red, Green and Blue) per pixel, this is not strictly true. HD video is a YUV 4:2:2 video format (also known as YCbCr) not a RGB format. The color is compressed. Instead of three color channels per pixel there are two: Cb (Chroma Blue) and Cr (Chroma Red) and a luminance channel Y. The green information is calculated mathematically. The color sampling ratio 4:2:2 refers to the fact that there is twice the luminance information as color information. Even a relatively lossless codec like Pro Res HQ is a 10 bit YUV 4:2:2 video format (although Pro Res 4444 is not). However, Pro Res is superior to most HD video codecs that are 8 bit YUV formats, and often using 4:2:0 or 4:1:1 color sampling ratios.
RAW video formats, in comparison, are strictly RGB video formats, using three 10 bit (or higher) RGB channels per pixel without color compression.
The second difference is whether the video is using linear or logarithmic digitizing.
Linear and log encoded video use two different data levels (or ranges) to store image data. In Resolve these ranges are described either as Video Levels for linear HD video, or Data Levels for log RAW video. Each video clip is identified on import as having either video or data levels but this identification can be changed manually through Clip Attributes in the Media workspace. The designation by the software should be correct. Nevertheless it is important to pay attention to these settings when you monitor the video on an external color correction monitor and when you export using the Deliver workspace. I have already mentioned how to use the video level setting on video output to the FSI color correction monitor.
Here is what you should concern yourself with in the Deliver workspace in Resolve to preserve the proper luminance on export:
There are three windows: Basic, Intermediate and Advanced. In the Basic window, the video or data level that Resolve has assigned to the clips is taken into account. In the Intermediate window you have a choice of keeping the Auto level (what Resolve has assigned) for the export or manually choosing Video or Data levels. You may
have to choose manually if you are combining different types of media, linear and log, on a timeline.
Use the following scenarios as a guide. You should always reimport your exported file into Resolve to compare luminance levels with the original graded timeline.
1. A DPX export from a timeline with RAW media:
The Basic setting is correct.
Intermediate Video/Data Level settings:
The Auto and Data setting are correct.
Video setting is incorrect.
2. Pro Res HQ export from RAW media
Basic setting is incorrect.
Intermediate Video/Data Level settings:
Auto and Video are correct.
Data is incorrect.
3. Pro Res HQ export from Pro Res Log Video (from Black Magic Cameras)
Basic is correct
Intermediate Export Level Settings:
Auto and Video are correct
Data is incorrect
4. Pro Res HQ export from HD Video (linear video: H.264, XDCAM, AVCHD, etc.)
Basic is correct
Intermediate Export Level Settings:
Auto and Video are correct
Data is incorrect
For reference, here is the description from the Resolve 11 Manual of Video and Data Levels (P.101):
To generalize, with 10-bit image values (with a numeric range of 0-1023), there are two different data levels (or ranges) that can be used to store image data when writing to media file formats such as QuickTime, MXF, or DPX. These ranges are:
> Video Levels: Typically used by Y’CBCR video data. All image data from 0 to 100 percent must fit into the numeric range of 64–940. Specifically, the Y’ component’s range is 64–940, while the numeric range of the CB and CR components is 64–960. The lower range of 4–63 is reserved for “blacker-than-black,” and the higher ranges of 941/961–1019 are reserved for “super-white.” These “out of bounds” ranges are recorded in source media as undershoots and overshoots, but they’re not acceptable for broadcast output.
> Data Levels: Typical for RGB 444 data acquired from digital cinema cameras, or film scanned to DPX image sequences. All image data from 0 to 100 percent is simply fit into the full numeric range of 0 to 1023.
References:
Wright, Steve. Digital Compositing For Film and Video, Third ed., Focal Press, 2010.