UNDERSTANDING THE TRANSFORM
The standard Canon2Alexa transform is divided into two components, the camera match transform and the output transform.
The first component or transform matches the cameras in linear space and outputs the image in log. It takes Canon Log 2/Canon Cinema Gamut footage from the C70 and C300 Mark III as the input, linearises the image, matches it to the Alexa, then outputs this match as ARRI LogC/Alexa Wide Gamut.
The second component is an output transform LUT that takes the ARRI LogC/Alexa Wide Gamut image and converts it to Rec709. This takes the flat desaturated log image, applies a LogC to Rec709 2.4 gamma curve, an Alexa Wide Gamut to Rec709 matrix and outputs an image with restored contrast and saturation that will look correct on your display.
WHY TWO TRANSFORMS?
The main reason is so that footage converted with Canon2Alexa can be integrated into existing colour grading workflows designed for the Alexa, or so that it can integrate alongside footage from the Alexa allowing the same look to be applied to both sets of footage with little or no modifications.
The key component of Canon2Alexa is the camera match transform. This is actually where the “Alexa look” lives. The output transform, while important, isn’t actually a large part of what gives the Alexa its distinctive look. In most Alexa productions, the default ARRI Rec709 output transform LUT used in camera, is discarded when it comes to grading. And all grading is carried out on the ARRI LogC/Alexa Wide Gamut image. Therefore using just the first component of Canon2Alexa places the Canon footage at an equal start point to real Alexa footage. From this point the Canon footage can be treated as if it was real Alexa footage. Any grade, PowerGrade, look or LUT designed for the Alexa can now be used on the Canon footage with little or no modification.
WHEN YOU MIGHT WANT TO USE A SINGLE COMBINED TRANSFORM
In some situations or workflows it might be necessary to use a single combined transform. A combined transform is useful for previewing the look on-set via a LUT as it would be impractical and unnecessary to use two transforms at this stage. It can also be useful to use a combined LUT at the edit stage so you can work with a more finished looking image. Combined in-camera/on-set LUTs can be found in the Canon2Alexa Rec709 In-Camera and On-Set LUTs folder. If you purchased the LUT version of Canon2Alexa, high quality combined LUTs for post production can be found in the Canon2Alexa Rec709 Post-Production folder.
IMPROVING THE STANDARD ARRI REC709 OUTPUT TRANSFORM
While the standard ARRI Rec709 LUT is probably the best of all camera manufacturer LUTs, it can still be improved on. As part of 2Alexa I have made some modifications to the standard ARRI LUT. These new LUTs have been modified for a more finished look with more filmic saturation. They were born from my research into film emulation. They allow for deeper, richer colours whilst avoiding over-saturation. The new LUT is called ARRI_LogC2Video_709_V2.cube. And a lower contrast wide dynamic range version is called ARRI_LogC2Video_709_WDR_V2.cube
CAMERA SIDE SETUP
As mentioned above, Canon2Alexa expects Canon Log 2/Canon Cinema Gamut as the input. Below are the other recommended camera settings for the C70 and C300 Mark III.
Gamma: Canon Log 2
Gamut: Canon Cinema Gamut
ISO: Use whatever ISO setting you need to achieve correct exposure, but I generally stick to ISO800.
Temp/Tint: Use whatever temp and tint you need to achieve the correct white balance.
Codec: I recommend using the highest quality codec for your media and frame rate combination. For high frame rates and most media this will be XF-AVC. For 30fps or lower on the C70 this will be RAW ST. RAW LT is only recommended when shooting high frame rates and when you must shoot in raw. In my tests, XF-AVC has less compression artifacts in high detailed scenes when compared to RAW LT.
EXPOSING YOUR IMAGE
Part of why great looking Alexa footage looks the way it does can be attributed to correct exposure. Correct exposure ensures deep rich colours, rich skin tones and correct contrast via optimal use of the sensor’s dynamic range. This means methods such as ETTR or deliberate overexposure should be avoided.
ETTR is mostly ok for cameras recording to 8bit codecs. But ETTR is unnecessary for cameras recording 10bits+ in high quality codecs like raw, Prores or any flavour of XF-AVC at high bitrates.
ETTR creates issues with differing noise levels and exposure not matching between shots in the same scene. This is due to the exposure being dictated by the brightest element in the scene. And if the brightest element in the scene disappears or changes brightness, ETTR dictates that you change your exposure. Which means shots from the same scene won’t match in exposure or noise level. What might be even worse, is that it’s possible to expose so far to the right that it is difficult to return the shot back to its natural contrast, as the majority of the shot lives in the upper linear portion of the log curve. This will result in a thin looking image that lacks depth and richness.
A much better method than letting highlights dictate exposure, is to expose for the middle value of the shot, AKA 18% gray. The theory here is that if you expose for middle gray on a sensor that has roughly an equal amount of stops above and below middle gray, you’ll be able to accurately capture the entire dynamic range of the scene with correct contrast, minimal or no clipped highlights and all at an acceptable noise level. And the exposure will be consistent between shots and scenes. With cameras like the C70 and C300 Mark III, it is possible to do this for most scenes.
FALSE COLOUR
The easiest method to expose for middle gray is to use false colour. With most false colour implementations, anything that is highlighted green is being exposed exactly at the middle of the sensor’s dynamic range. Along with green there is usually a secondary colour such as pink that is 1 stop brighter than middle gray. This is to help you expose skin. For most shots featuring people, I usually place green towards the shadow side of the skin, and place pink on the lit side of the skin. Any skin tones between those colours will be right in the sweet spot. For darker complexions it’s possible place green further towards the lit side of skin to avoid overexposure.
For scenes that don’t feature people, it’s a bit more flexible. Generally I place green on or near the focus of the scene. But paying close attention to the brightness of that focus. Placing green on a white polar bear in snow will result in an underexposed image. Here you need to use judgement and experience to gauge correct exposure.
HOW DO YOU KNOW WHEN YOU HAVE CORRECT EXPOSURE?
If you can take your footage into Resolve and apply Canon2Alexa and without any adjustments the skin exposure looks correct, you can simultaneously see detail in the shadows and highlights, and you have no urge to adjust exposure, then your exposure is spot on.
For some projects shot by talented DPs, I can grade the entire project without adjusting exposure. Thats the level of accuracy possible with false colour.
INSTALLATION AND USAGE
IN-CAMERA AND MONITOR LUTS
The in-camera and monitor LUTs can be found in the Canon2Alexa Rec709 In-Camera and On-Set LUTs folder.
I prefer loading the LUTs onto an external monitor. Loading them within a custom look file will bake the LUT into the footage when shooting non raw formats. This will obviously then limit what can be done in the colour grade.
LUT VERSION RESOLVE INSTALLATION
Extract the zip file, open Resolve and bring up Project Settings. Navigate to the Color Management tab on the left and find the Lookup Tables section. Click the Open LUT folder and copy over the Canon2Alexa LUT folder. If you also want the all-in-one Rec709 versions you can also copy the Canon2Alexa Rec709 Post-Production folder as well. Navigate back to Resolve and click the Update Lists button. The LUTs will now be accessible inside Resolve.
RESOLVE COLOR MANAGEMENT SETUP
In the Color Management tab make sure the Color Science is set to the default of Davinci YRGB and the Timeline color space is set to Rec.709 Gamma 2.4. If you’re on a Mac and usually set your timeline colour space to Rec709-A, leave the Timeline color space as Rec709 Gamma 2.4 and set Rec709-A as the gamma tag in the delivery page.
The PowerGrade is currently not compatible with the Davinci YRGB Color Managed workflow. I’ll look at creating an RCM compatible version in the future.
USING THE LUTS IN RESOLVE
There are two versions of the Canon2Alexa LogC transform LUTs. Two LUTs are required as there is a slight difference in the colours when raw files are decoded using Resolve’s raw decoder. If your footage is Raw ST or Raw LT, you are grading in Resolve and you have the Decode quality set to Full res - Resolve in the raw tab like the picture below (which is the default setting), then use Canon2Alexa_LogC_Resolve_Raw_Decode.cube.
If your footage is shot in any of the XF-AVC or MP4 codecs use the standard Canon2Alexa_LogC.cube LUT. If your footage is Raw ST or Raw LT and you are grading in an application other than Resolve, you should also use Canon2Alexa_LogC.cube.
This is the basic node structure I recommend for the LUTs.
Node 01 is noise reduction. Whether noise reduction is needed depends on the footage. Normally I use NeatVideo, but Resolve’s noise reduction is actually very good. When using Resolve’s noise reduction I generally use Spatial NR with Mode set to Faster, Radius set to Small. Then under Spatial Threshold I unlink Luma and Chroma and leave Luma at 0.0 and set Chroma to around 3.0-8.0.
Node 02 as mentioned above, if your footage is shoot in any of the XF-AVC or MP4 codecs, place the Canon2Alexa_LogC.cube camera match transform LUT in this node. If your footage is Raw ST or Raw LT and you’re using the Resolve Canon Raw decoder (which is the default setting), place Canon2Alexa_LogC_Resolve_Raw_Decode.cube in this node. Avoid making any changes prior to this node as it will affect the accuracy of the camera match transform.
Node 03 is used for image balancing, generally using the offset colour wheel, offset master wheel and contrast/pivot. The idea here is to balance the image and correct for shot to shot differences. For colour balancing use the offset colour wheel. Use the master offset wheel (slider under the offset colour wheel) for exposure changes and use contrast and pivot to dial in the contrast for your image. Working in this node and onwards is the same as working on real Alexa footage.
The gap between Node 03 and 04 this is where creative grading takes place. You can use any tool or technique here. Use whatever is needed to achieve the look you are after.
Node 04 is used for the output transform LUT. This node handles the conversion of ARRI LogC/Alexa Wide Gamut to Rec709. Select a LUT from the Canon2Alexa/Alexa Rec709 folder or any other LUT designed for ARRI LogC. You can also use a Colour Space Transform node in place of this node to handle the conversion for extra flexibility. Or you can remove this node completely and append one of my YRGB print film emulation PowerGrades.
USING THE LUTS WITH OTHER SOFTWARE
The are two methods to use the Canon2Alexa LUT transforms in NLEs such as Adobe Premiere.
The easiest method is to use the all-in-one LUTs included with the LUT version of Canon2Alexa. Apply one of the LUTS from the Canon2Alexa Rec709 Post-Production folder, then grade underneath this LUT. The way I would set this up in Premiere is to create an Adjustment Layer that sits over the entire timeline. Add a Lumetri effect to the Adjustment Layer and select one of the all-in-one LUTs. Add another Lumetri effect to any clip that needs grading/correction. Corrections to each clip are now happening under the main Adjustment Layer LUT.
The second method is slightly more involved but it’s more accurate and closer to working with real Alexa footage. It also gives you more flexibility as you can choose to use any output transform LUT made for the Alexa. The way to set this up in Premiere is to apply a Lumetri effect to a clip, then in the Input LUT field select the Canon2Alexa_LogC.cube LUT from the Canon2Alexa/Canon2Alexa LogC folder. Then copy this Lumetri effect to every clip in your timeline. Create an Adjustment Layer that sits over the entire timeline, add a Lumetri effect to it and select any LUT from the Canon2Alexa/Alexa Rec709 folder. You can also select any LUT made for the Alexa. Any corrections made to the Lumetri effect at the clip level are happening under the main Adjustment Layer LUT.
POWERGRADE INSTALLATION AND USAGE
POWERGRADE INSTALLATION
To install the PowerGrades in Resolve, navigate to the PowerGrade 1 stills album in the Gallery. Adding PowerGrades to this album will allow you to access the PowerGrades from any project and it will also mean you only have to import them once. Once the album is selected, second mouse button click in the gallery window background pane and select Import. Navigate to the extracted zip file location and select the .drx file(s) and click import. If it doesn’t let you select a file, you might need to click the options button at the bottom of the import window and then select .drx from the list, then try again.
Along with the PowerGrade you’ll need to install the ARRI LogC to Rec709 conversion LUTs. Follow the LUT installation instructions above and copy the Canon2Alexa Rec709 LUTs folder to Resolve’s LUT folder. If you do this incorrectly, your footage will remain as flat log gamma after applying the PowerGrade. You will also get a missing LUT error message.
USING THE POWERGRADES
Once installed, the PowerGrades can be applied to the selected shot by middle mouse button clicking on the thumbnail still in the gallery. Or with a second mouse button click on the thumbnail and then selecting Apply Grade from the menu.
Similar to the LUT version of Canon2Alexa, there are two versions of the Canon2Alexa PowerGrades. Two PowerGrades are required as there is a slight difference in the colours when raw files are decoded using Resolve’s raw decoder. If your footage is Raw ST or Raw LT, and you have the Decode quality set to Full res - Resolve in the raw tab like the picture below (which is the default setting), then use Canon2Alexa Resolve Raw Decode PowerGrade.
If your footage is shot in any of the XF-AVC or MP4 codecs use the standard Canon2Alexa PowerGrade. If your footage is Raw ST or Raw LT and you have the Decode quality set to Full res - Canon, you should also use the Canon2Alexa PowerGrade.
This is the structure of the PowerGrade.
Node 01 corrects the individual RGB channels via the RGB curves.
Node 02 is a Full to Video Levels conversion. For some reason any CST conversion that involves the Canon Log 2 curve also seems to perform a Video to Full Levels conversion as part of the transform. This is very strange behaviour and is completely different to any Canon Log 2 to ARRI LogC or Linear conversion performed by LUTCalc or Lattice. Until BMD corrects this behaviour, leave this node on.
Node 03 is a CST node that transforms Canon Log 2 to linear.
Node 04 is the 3x3 matrix and the core of the camera match transform. This matrix lives in the RGB mixer and it aligns the Canon colours to match the Alexa.
Node 05 is for accurate exposure adjustments using the master gain wheel (the slider under the main gain colour wheel). Doubling the gain from 1.00 to 2.00, 2.00 to 4.00 etc increases the exposure by precisely 1 stop. It doesn’t need to be whole rounded values, any value that is doubled is an increase of one stop. In the same way halving from 1.00 to 0.50, 0.50 to 0.25 etc decreases the exposure by 1 stop. And again it’s any value, not just whole rounded values. You can also colour balance in this node by using the gain colour wheel, which results in more accurate and cleaner results than using the offset wheel and sometimes even better results than the temp/tint controls when working with raw footage. Do not use any tool other than master gain wheel and colour gain wheel in this node as it will produce incorrect results.
Node 06 is a CST that transforms the image from linear into ARRI LogC. If you have any colours that are clipping or causing out of gamut issues, try enabling Gamut Mapping in the CST in this node. This node is the final node of the camera match transform.
The Gap between Node 06 and 07 is where creative grading takes place. You can use any tool or technique to achieve the look you are after. Grading at this point is the same as grading real Alexa footage.
Node 07 is the output transform. It handles the conversion from ARRI LogC/Alexa Wide Gamut to Rec709. This node can be replaced with any LUT in the folder. Or any other LUT designed for ARRI LogC. You can also use a Colour Space Transform node in place of this node to handle the conversion for extra flexibility. Or you can remove this node completely and append one of my many YRGB print film emulation PowerGrades.
Much the same as the LUT version, avoid making any adjustments prior to the first node as this will affect the transform’s accuracy.
Following my standard Alexa workflow, for most grades I remove Node 06 and replace it with a Print Film Emulation (PFE). Either a real PFE LUT or one of my PFE PowerGrade rebuilds.
With or without the PFE, all grading should take places after Node 05 but before the output transform.
Similar to the LUT version of the PowerGrade, I pretty much always add noise reduction in a node before Node 01. I mostly use NeatVideo, but I sometimes use Resolve’s noise reduction. For noise reduction settings check the LUT section above.
CANON2ALEXA RCM POWERGRADE
The Canon2Alexa RCM PowerGrade allows you to match the C70/C300 MKIII to the Alexa when using the Resolve Color Managed workflow. It should produce better results when using RCM Print Film Emulation PowerGrades.
Node 01 corrects the individual RGB channels via the RGB curves.
Node 02 is a Full to Video Levels conversion. For some reason any conversion that involves the Canon Log 2 curve also seems to perform a Video to Full Levels conversion as part of the transform. This is very strange behaviour and is completely different to any Canon Log 2 to ARRI LogC or Linear conversion performed by LUTCalc or Lattice. Until BMD corrects this behaviour, leave this node on.
Node 03 is the 3x3 matrix and the core of the transform. This matrix lives in the RGB mixer and it aligns the C70/C300 MKIII colours to match the Alexa.
A node after Node 03 is where I balance the image and make exposure adjustments. This is best done in the HDR Color Wheels panel, using the Global Colour Wheel for balancing out colour casts and the Exp control to adjust exposure. You can also append other RCM PowerGrades such as the RCM PFE PowerGrades after this balance/exp node.
To use the RCM PowerGrade, set the Color science to Davinci YRGB Color Managed, Colour processing mode to HDR Davinci Wide Gamut Intermediate. And then set the Output color space to match your deliverable’s colour space.
If your footage isn’t a flavour or raw, you will need to manually tell Resolve the gamut and gamma of your media. To do this navigate to the media pool, select all your clips, second mouse button click on one of the shots, select Input Color Space, then select the gamut/gamma of your footage. For Canon C70/C300 MKIII footage this will be Canon Cinema Gamut / Canon Log 2.
QUESTIONS, COMMENTS, SUPPORT
For questions, comments or if you need LUT/PowerGrade support feel free to contact me at support@juanmelara.com.au