A new firmware with a kind of cryptic image related update. What does “Optimized color shading in for improved footage quality” even mean?
Do you any any guess, please leave it the comments below. Geert-Jan has an interesting insight and Eric Clayborn pointed to a particular issue that seems about right. Read below.
Release notes for v01.03.0800.
Download the .pdf here
- 2017.06.08 v01.03.0800 v4.1.0
Updated No Fly Zone Management.
For increased safety, the flight is restricted to a height of 30 m and distance of 50 m when not connected or logged into the app during flight, including DJI GO 4 and all apps compatible with DJI aircraft.
Added gimbal control in Fixed-Wing.
Optimized color shading in for improved footage quality.
Fixed minor drift issue in Portrait.
This firmware is also an update for DJI Goggles:
Added value adjustment gain for Head Tracking.
Added screen cut function for two fingers.
Added support for Micro SD card hot plug/unplug. Reduced HDMI input latency.
UPDATE 1 – Test with an existing LUT on a Colorchecker photo.
ICARUS LUTs behave just the same as they did before.
This update doesn’t relate to white balance or the overall color balance. From these measurements, ICARUS remais spot on when it comes to balancing the neutral patches on this Colorchecker.
Here we see the RGB values with a ICARUS V.2.1 DJI MAVIC PRO D-LOG Sunny WB – ETTR_64 point LUT applied to a D-LOG (+1;-1;0) Sunny WB frame.
UPDATE 2 – The much hated warm spot in the middle of the frame and how ICARUS can help
MAVIC pilot and color grader Eric Clayborn raised a very interesting point that relates to image shading, or shade in. In this blog post users refer to a red spot in middle of DJI Mavic footage. Lookup tables work on the whole frame and never with specific spatial coordinates. Also, this magenta spot is too mild to fit the category of a very distorted color that would be corrected whenever it appeared anywhere in the frame. So, how can ICARUS help with an uneven spot? By observing some samples, we can assume that this firmware update diminished the warm spot at the center of the frame, thus placing the magenta spot in the same ballpark as the overall magenta cast in sunny white balance. And here is where ICARUS can make a huge difference, if used correctly. So, here we go:
First we have the original clip in D-Log. The magenta spot is visible in the center.
Then, the same clip corrected with an ICARUS LUT.
Followed by an exaggerated correction in the same node that contains the LUT. Note how the magenta spot is still present.
And ending with an interesting conclusion. When the LUT is applied before the node with all the exaggerated contrast and saturation, we get a perfectly neutral image and that magenta spot is gone, this time replaced by a mildly dark spot.
This proves that not only does ICARUS fix part of the warm spot issue after the .800 firmware but, again, it should be used as a primer at the beginning of your grade.
What about a JPEG photo
JPEGs are output through the same pipeline as the video signal so we’d expect the same results. Left is native, right is pushed in Lightroom. Again the warm spot in the middle. Is it weaker than before?
Here are the links to the original frames.
Before and after
JPEG photo with an ICARUS LUT applied
In this example we see the same jpeg frame with the ICARUS V.2.1 DJI MAVIC PRO D-LOG Sunny WB – Flat_64 point applied. Immediately we can see a different story in Photoshop. We have a neutral white across the whole surface, including the center. This is expected since the MAVIC’s compressed H.264 and JPEG output in sunny white balance lacks a constant value on the green channel across the whole range, making the picture look magenta.
Click on the picture to open in a new tab.
DNG photo. Same, but different
Here we see a DNG file converted in Lightroom. Again, the first frame shows a realistic depiction of the white patch with appropriate white balance and luminance levels as seen in the color picker. The second frame shows the same image with exaggerated curves and contrast. We deduct the same as with the video file, there is still some sort of color shading at the center of the frame. The difference this time is that by using DNG we are getting a file that is closer to the sensor’s native readout, thus potentially exposing native hardware flaws that can’t be compensated for by the Ambarella A9A0 SoC that powers the MAVIC’s image pipeline.
Case and Point
1. This seems like a hardware issue with the sensor/lens combination
2. Yes, the hotspot issue is still there after the .800 update although by looking at images in various forums, we can assume that it has been reduced. If you have a .700 MAVIC, can you make a before/after of an evenly lit white surface in D-Log (+1;-1;0) Sunny?
3. It is a magenta hotspot and although DJI couldn’t reduce the darkening at the center of the frame, at least it made it match the overall magenta cast across the picture that has been talked about in other blog posts.
4. And this makes it possible for the ICARUS LUTs to reduce and eliminate the color cast completely, given that they’re applied before any subsequent color grading. We end up with a slightly darker area at the center of the frame that can be easily compensated for with a mask.
5. There is no need to update ICARUS due to this firmware update.
Again, if you need a test run of ICARUS, send me a message with some original files in D-LOG or D-Cinelike and I’ll get back with the results.