Final Cuts: Of Colour Corrections, Last Edits, and the Results of My Learning
As I blogged in my last blog entry, I ended my previous round of editing with a couple of issues surrounding the Latitude Tests and colour correction, which I tried to tackle this week along with finalising the edit by cleaning up, adding music, exporting and conducting test runs on the big screen. For this, I had a total of 20 hours of editing time reserved.
But let me start at the beginning.
Tuesday, 19 November 2019
On Tuesday, I was finally able to correct my colour correction. As you might remember from my last blog entry, one problem I faced was the fact that DaVinci Resolve 16 did not have a preset for the Source Gamma of the Blackmagic Cinema Camera 2.5k. Having played around with the Blackmagic Film Design preset, I quickly encountered issues with my overexposed footage, as I started to get colour artefacts, rendering the overexposed parts of the image blue:
Because of this, I then decided to test the Source Gamma of the Blackmagic URSA Mini Pro 4.6k to test whether that could resolve the issue. Et voilà, the colour artefact vanished:
After that, I ventured into colour correcting the clips I took for my ISO/final lighting test. However, since I did not use a colour chart for this particular test, I had to manually colour correct the clips, which I did by adjusting the colour wheels for the highlights, midtones and shadows until I was satisfied with the result. I then put this colour correction as a preset for all the remaining clips to ensure that the only difference between them would be the difference in ISO – allowing for comparison of the camera’s performance under low light conditions. I then moved back to Avid to edit all the colour corrected clips of the ISO test.
Since I wanted to keep the results of my camera testing as clear as possible without diluting them by too much manipulation – and since time was of the essence as of this point – I decided against colour grading the footage and instead proceeded with the editing of the colour corrected Latitude tests:
Wednesday, 20 November 2019
Since I began to become blind on my own editing on Tuesday, I decided to have another round of checks on Wednesday morning to ensure that I had not overseen anything. I then cleaned up loose ends and corrected some of the titles before I moved onto editing the soundtracks:
After that was finally done, I was looking at a neat collection of clips to export, which I did in bulk to save time and ensure that all was done to the same settings:
I then exported the clips onto the desktop of my student account for easy access on the computer in the film seminar room, in order to have a test run on the projector and check how the playback would be on a big screen in general:
Watching the clips back on a projector finally allowed me to fully analyse all the footage and camera tests in detail and to full extent. While I naturally analysed bits and pieces along the entire process from capture to editing and colour correction, it was obvious that viewing everything in context and on a big screen – in contrast to a small LCD camera screen or the monitors in the Edit Suite that might mask certain issues – would be more beneficial.
As such, and without going into too much detail – most of this will be covered in my presentation anyway – here are a couple of my observations I realised when analysing my footage:
My Analysis – The Results of My Learning
Don’t trust the focus peak option on the URSA! As can be gleaned from the overexposure clips of the URSA, and although I made sure that I focussed on Hannah’s eyes, lips, and bridge of the nose, as well as the grid of the colour chart, the footage is extremely soft up – only showing the colour chart in focus – until I manually refocused at +2 stops. After a bit of research, I found out that the focus peak function of the URSA indicates focus not based on the actual focus points, but based on contrast within the image. The clearer the line(s) between a dark and a bright area of the image, the more likely it will be for the camera to pick that border as a reference frame for focus.
When overexposing footage on the URSA (stopping up), the depth of field gets narrower as expected. The focal point of the URSA however remains steady throughout all tested stops of overexposure.
When underexposing the footage on the URSA (stopping down), the depth of field expands, as expected. The focal point of the URSA however remains steady throughout all tested stops of underexposure.
When overexposing footage on the BMCC (stopping up), the depth of field gets narrower as expected. However, in contrast to the URSA, the focal point of the BMCC shifts forwards slightly, towards camera.
When underexposing footage on the BMCC (stopping down), the depth of field expands as expected. The focal point of the BMCC however remains steady.
The sensor sizes (as well as the colour spaces) on the BMCC and the URSA are different, reproducing different results.
When colour correcting with the colour chart, DaVinci Resolve 16 does not offer a Source Gamma Preset for the BMCC 2.5k, making it a bigger task to colour correct footage, requiring you to
Shadows on a colour chart are not your best friend. Whilst I was able to find a shadow-free reference frame in all clips but three to take a colour correction from, it unnecessarily overcomplicates matters to no end and should not be happening at a professional camera test.
Although both cameras do record 10-bit of colour space when shooting in ProRes 422, the URSA does show significantly more banding than the BMCC, especially in colour corrected footage with lower ISOs. At the same time, the colours of the URSA do look much more vibrant than the colours of the BMCC.
Straight off the camera and compared to the URSA, the BMCC’s greens are flatter, the blues more toned and the reds rather orange, whereas the URSA’s greens are stronger, the reds more towards magenta, and the blues more vibrant.
Although the BMCC footage straight off the camera seems to be safe to be underexposed by -2½ and overexposed by +1½ stops, the corrected footage still reveals grain at -2 stops and clip at +1 ½ stops. This would effectively mean that an underexposure by -1½ and an overexposure by +1 stop would technically still be safe.
Although the URSA footage straight off the camera seems to be safe to be underexposed by -2½ and overexposed by +1½ stops, the corrected footage still reveals grain at -2½ stops and clip at +1 ½ stops. This would effectively mean that an underexposure by -2 stops and an overexposure by +1 stop would technically still be safe.
Furthermore, I could neither confirm nor deny the claim that the URSA underexposes its footage by -2 stops.
Ooops, guess I did go well into detail regardless! Well, at least that is the condensation of my analysis. I then ended the day by re-exporting one clip in which the sound levels were not matching up with the other clips and putting everything together on my student desktop, leaving the hand-in of the individual clips into the Minor Project Hand-In folder for Thursday.