Author Topic: Introduction and showing my open-source workflow  (Read 5205 times)

GrumpyDeveloper

  • Newbie
  • *
  • Posts: 3
  • Developer, mostly happy, sometime grumpy :)
    • View Profile
    • By personal crazy projects
Introduction and showing my open-source workflow
« on: December 06, 2022, 09:42:50 AM »
Hi,
I'm a new owner of a used Chronos 2.1.
Since I could not find a good description of the workflow you use, I decided to explain my developed, pure open-source workflow for other beginners
(like me) or even some pros  ;) :
Feel free to correct mistakes and give additional hints.
Let's start with a link to my first public video:
https://www.youtube.com/watch?v=prXfGrHT3OE
The descriptions and overlays are in German, but I'm sure you get the point even if you don't understand the language.

This was very challenging, but a good example of the workflow I set up.

Let's start with the Scene setup.
I wanted (and needed) to film with 4300FPS. That meant a LOT of light, also remember the Inverse Square Law, so place the light source
as close as possible. I had a 200W LED around 15cm away, barely outside the view.
Lens choice is a critical one, depending on the subject. Physically close with wider lens or further apart with more tele-lens. All affects depth of field.
There are good calculators for this online. Also how big can/should the aperture be, all a matter of subject to film.
I chose a 70mm lens at F/4.

Camera settings:
A few words before, I used the newest firmware (0.7 at the time), but made a few changes. In the end, it's an embedded linux device,
so attach a network cable and ssh into it. You can find the login via search, not sure if I can post them here.
Run the updates via "apt update" and "apt upgrade", the update over the ui does not work, because the krontech certificates are messed up
at the time of doing this. Not sure if fixed by now.
Also when in there, add an ntp client so you don't have to manually set date/time. My camera is bad at tracking the current time.

Storage:
This one took me take a while, but these are my results:
Storage via network share NFS, SSD might be faster, but you need to unplug, plug into the computer a.s.o. This takes longer than NFS share
to the target computer.
Storage type:
I store tiff and cameraDNG. First one can be used in the image viewer of my choice (kwenview) to quickly find the "real" start and "end" of the scene.
I personally find the UI too clumsy so I just tend to be on the side of caution.
The raw formats (a single file) may be faster, but more difficult to find the start and end. Raw tiff export created grayscale images, not sure why.

Process with shooting the scene, you mileage will vary.
Store both tiff and cameraDNG.

At this point you should have a few thousand images on your target computer.
As mentioned, first find the start and end, move the used images into an own folder.

Postprocessing:
I fire up g'mic-QT with one of the images to find the processing steps. You can do it with gimp or similar, but g'mic is so extreme powerful (584 filters in current version)
and can be run easily in batch processes.
For the above mentioned video I had to do:
- rotate image 90°
- crop
- apply a brightness curve, the images were way too dark
- resize
- denoise (AI based) 3x for best results
- sharpen
and this 1225 times.
So here was the command, sorry windows users:
Code: [Select]
ls dng/*.dng | sed -s "s/.*\///" | sed -s "s/\.dng//" | xargs -I FILE bash -c 'echo FILE; nice -n20 gmic input dng/FILE.dng -rotate 90 -crop 0,250,359,889 -apply_curve 1,10,0,600,50,3000,110,65535,255 -resize 608,1080 -denoise_cnn 3 -denoise_cnn 3 -denoise_cnn 3 -sharpen 200 output out/FILE.pngIt probably can be done on windows and mac, I'm just not using those.
The built in editing tools of your video editor have no chance against a properly tunen g'mic process. Not even the expensive ones.
May be more interactive, and easier to click around, but quality and flexibility is worse.

After this, you get the raw image series for use in your favorite video editor.
If there is interest, I can post a follow up for the remaining workflow (a custom script for subtitle generation, kdenlive, audacity)

Cheers Grumpy

edit: fixed firmware version.
« Last Edit: December 06, 2022, 03:31:04 PM by GrumpyDeveloper »

ivandan

  • Newbie
  • *
  • Posts: 38
    • View Profile
Re: Introduction and showing my open-source workflow
« Reply #1 on: December 07, 2022, 03:49:55 PM »
It looks easy to me
1. Production: Take a shot (used 1000fps, 0db, DNG on SSD, standard latest firmware and color scheme, have not yet figured out how to work with colorchecker)
2. Postproduction in Davinci Resolve (drag and drop files into the program, custom secret colorgrading and processing techniques film development, 4x renders, i get paid for this)

https://vimeo.com/745084986
https://vimeo.com/767261235/71c0f32d7d

Visualtec

  • Newbie
  • *
  • Posts: 36
    • View Profile
Re: Introduction and showing my open-source workflow
« Reply #2 on: December 08, 2022, 04:27:58 AM »
Thanks for sharing some interesting things there.

Nikon1

  • Hero Member
  • *****
  • Posts: 983
    • View Profile
Re: Introduction and showing my open-source workflow
« Reply #3 on: December 08, 2022, 02:01:08 PM »
[..] looks easy to me [..]
 

 Damn, and i thought i used a ton of Modifiers on my Grading and Color Correction... :o