More Photo and Video fun stuff

We come from…getting to know what ffmpeg is for benchmarks to compare CPU performance.

Wait, there is even a repo for video stuff!

Video Edition Repository

Quick videos as a code from Action Cams

And now there is a chance to do a photo & video recap.

Video Editing

In that way the acelerometer information will be captured in the mp4 file.

CLI Tricks

There is people outthere doing cool things with ffmpeg: https://blog.bartzz.com/create-stunning-timelapses-with-ffmpeg/

Transfering Files

mkdir -p "$HOME/Desktop/CAM1-$(date +%m-%d)" && rsync -avP --include='*.MP4' --exclude='*' . "$HOME/Desktop/CAM1-$(date +%m-%d)"

Videos with FFMPeg

Just put your video parts together / silenced / with a given .mp3 audio:

#Cut between two second or from a second to the end
#ffmpeg -i DJI_20250515200111_0003_D.MP4 -ss 10 -to 90 -c copy output.mp4
#ffmpeg -i DJI_20250515200111_0003_D.MP4 -ss 91 -c copy output_.mp4
ffmpeg -i "$(ls *.MP4)" -ss 26 -c copy output_.mp4

#Last 52s
ffmpeg -sseof -52 -i DJI_20250520200235_0006_D.MP4 -c copy output_last_52s.mp4

## For Tinkering with more video parts
ls *.MP4 | sed "s/^/file '/; s/$/'/" > file_list.txt #add .mp4 of current folder to a list

#Generate a video from few parts
#ffmpeg -f concat -safe 0 -i file_list.txt -c copy output_video.mp4
ffmpeg -f concat -safe 0 -i file_list.txt -c:v copy -an silenced_output_video.mp4 #silenced video
#ffmpeg -i output_video.mp4 -filter:v "setpts=PTS/4" -an fast_output_video.mp4 #x4

#ffmpeg -stream_loop -1 -i "AETHER - Density & Time.mp3" -i silenced_output_video.mp4 -c:v copy -c:a aac -shortest output_with_song.mp4
ffmpeg -stream_loop -1 -i "output.mp3" -i silenced_output_video.mp4 -c:v copy -c:a aac -shortest output_with_song.mp4

### 🎵 Music by...

#from second 90 of the video, give me 1fps
ffmpeg -i DJI_20250116072852_0036_D.MP4 -vf "select='gte(t\,90)',fps=1" -vsync vfr frame_%03d.png
ffmpeg -i DJI_20250116072852_0036_D.MP4 -vf "select='gte(t\,90)',fps=1" -vsync vfr frame_%03d.jpg

ffmpeg -i DJI_20250116072528_0035_D.MP4 -vf "select='between(t,90,105)',fps=1" -vsync vfr frame_%03d.png

4. Applying custom luts with ffmpeg:

#https://video.stackexchange.com/questions/21483/apply-custom-lut-via-ffmpeg
ffmpeg -i "Input.mov" -vf lut3d="ARRIP3D65PQ108-33.cube" -s 1920x1080 -c:v dnxhd -pix_fmt yuv422p -b:v 120M DNxHD_for_Editing.mxf

Shorts

While 1080p (Full HD: 1920x180) is a very common and recommended resolution for YouTube, you can definitely upload videos in other resolutions, both higher and lower.

• Frame Rate: Upload your video in the same frame rate it was recorded. Common frame rates include 24, 25, 30, 48, 50, and 60 frames per second.

• Aspect Ratio: The standard aspect ratio for YouTube is 16:9. While other aspect ratios are supported, 16:9 generally provides the best viewing experience across different devices.

4K UHD (Ultra High Definition): This is the most common 4K resolution for televisions, monitors, and streaming services. Its dimensions are 3840 pixels wide by 2160 pixels high (3840 x 2160). This resolution has exactly four times the number of pixels as 1080p (1920 x 1080).

How about adding TTS generated audio to shorts?

1. Generate the AI Audio with TTS:

soxi -D audio_reply.wav #video duration

2. Get your short content transfered:

mkdir -p "$HOME/Desktop/CAM1-$(date +%m-%d)" && rsync -avP --include='*.MP4' --exclude='*' . "$HOME/Desktop/CAM1-$(date +%m-%d)"

From SD card I get transfers at 68MB/s and from the internal ~200MB/s (when using an external nvme case usb-c)

3. Make 175s pieces max:

#Shorts (limited to 3min or generated audio duration)

#As per audio duration, which is passed to the CLI
ffmpeg -i DJI_2025abc123.MP4 -t $(soxi -D audio_reply.wav) -c copy cut_output.mp4
ffmpeg -i "$(ls *.MP4)" -t $(soxi -D audio_reply.wav) -c copy cut_output.mp4

#As per hardcoded duration
#ffmpeg -i DJI_2025abc123.MP4 -t 175 -c copy output.mp4

4. Extract the audio of the .MP4 piece, and combine it with the ai generated audio:

#Take the cut video and combine its audio with the AI generated
ffmpeg -i cut_output.mp4 -i audio_reply.wav \
-filter_complex "[0:a:0]volume=0.6[vid_audio];[1:a:0]volume=1.4[reply_audio];[vid_audio][reply_audio]amix=inputs=2:duration=longest[aout]" \
-map "[aout]" -c:a libmp3lame output.mp3
# ffmpeg -i DJI_20250511143231_0004_D.MP4 -i audio_reply.wav \
# -filter_complex "[0:a:0][1:a:0]amix=inputs=2:duration=longest[aout]" \
# -map "[aout]" -c:a libmp3lame output.mp3

# ffmpeg -i DJI_20250511143231_0004_D.MP4 -i audio_reply.wav \
# -filter_complex "[0:a:0]volume=0.6[vid_audio];[1:a:0]volume=1.4[reply_audio];[vid_audio][reply_audio]amix=inputs=2:duration=longest[aout]" \
# -map "[aout]" -c:a libmp3lame output.mp3

#with just one video in the folder:
ffmpeg -i "$(ls *.MP4)" -i audio_reply.wav \
-filter_complex "[0:a:0]volume=0.6[vid_audio];[1:a:0]volume=1.4[reply_audio];[vid_audio][reply_audio]amix=inputs=2:duration=longest[aout]" \
-map "[aout]" -c:a libmp3lame output.mp3

5. Put back the combined audio to the .mp4:

#For a Quick Short:
ffmpeg -i cut_output.mp4 -i output.mp3 -map 0:v -map 1:a -c:v copy -c:a libmp3lame -q:a 0 -shortest final_output.mp4

You can also take a 1080p part from a 4K video withour rencoding:

ffmpeg -i DJI_20250518172824_0001_D.MP4 -c copy -bsf:v hevc_metadata=crop_left=960:crop_right=960:crop_top=0:crop_bottom=1080 output_horizontal_1080p_no_encode.mp4

AI Powered shorts

A friend told me recently to have a look to the following to create interesting youtube shorts:

• Suno.ai for audio/songs…
• Klingai.com for video generated from images/text
• Ilevenlabs to create TTSpeech
  • Thanks to GitPodcast, I saw that its also possible to do with Azure AI Speech

Whatever you do, dont forget to add proper video description

Its kind of what these channels do!

Telemetry

Initially I tought that would be possible only with the GoPro…

• https://goprotelemetryextractor.com/cycling-stats-video-overlay?affi=safabrian#download_0
• https://jalcocert.github.io/JAlcocerT/tinkering-telemetry-trackdays/

But I was wrong, action cameras, including the oa5pro, have built in acelerometers (just that not all of them have GPS’s):

#choco install exiftool
#exiftool -ver
exiftool -ee DJI_20250518182847_0015_D.MP4 #working when recorded W and no RS!

Adding Telemetry to MP4s

How to Telemetry Overlay a Video

I tried to bypass this with: https://jalcocert.github.io/JAlcocerT/polar-data-python-analysis/

and with https://jalcocert.github.io/JAlcocerT/blog/tinker-phyphox/

Then the geospatial analysis came…https://jalcocert.github.io/JAlcocerT/geospatial-data/

Kdenlive

I was tinkering with KDEnlive sometime ago, with the x300…

flatpak install flathub org.kde.kdenlive

This video was very helpful to understand additional KDE tricks:

Have rediscovered KDEnlive for LUT (.cube files) color grading application!

I personally like the CINEMATIC.cube that KDE brings:

#find / -name "*.cube" 2>/dev/null
#cd /var/lib/flatpak/app/org.kde.kdenlive/x86_64/stable/1e36391e875577fd4cd61f5a02e7e52d1cf22598a625d235d49546fd39b77c0c/files/share/kdenlive/luts
cp CINEMATIC.cube /home/jalcocert/Desktop

OA5Pro

Action Cam Review

KDenLive with OA5Pro

Editing video with kdenlive

Those LUT .cube files can potentially be applied via FFMPEG CLI:

# ffmpeg -i DJI_20250518175448_0005_D.MP4 -ss 0 -to 10 -c copy output_first_10_seconds.mp4
# ffmpeg -i output_first_10_seconds.mp4 -i ./CINEMATIC.cube -filter_complex "lut3d" -c:a copy output_first_10_seconds_cinematic.mp4

Gyroflow

I already had a look to Gyroflow on the last video post.

• https://github.com/gyroflow/lens_profiles/tree/main/DJI

And it can also be done with gyroflow-CLI:

gyroflow-cli --input DJI_20250518182847_0015_D.MP4 --output stabilized_video.mp4

Filters

ND Filters

A Neutral Density (ND) filter is essentially a sunglass for your camera lens. It’s a filter made of darkened glass or resin that reduces the amount of light entering the camera sensor without affecting the color or hue of the scene.

How they work:

• ND filters are designed to reduce the intensity of all wavelengths of light equally. This means they dim the overall brightness of the scene without introducing any color casts (in theory; some cheaper ND filters might have slight color shifts).
• They work by absorbing or reflecting a certain percentage of the incoming light.

Photo Editing

Spending a lot on smartphone with cool cameras to…use snapseed?

Come on… whats next?

Uploading to social media in low resolution?

Lets do some CLI Photo tinkering:

file libreportfolio.png
#libreportfolio.png: PNG image data, 2048 x 2048, 8-bit/color RGBA, non-interlaced

convert libreportfolio.png -resize 1200x630^ -gravity Center -extent 1200x630 -quality 85% libreportfolio-og.jpeg
convert libreportfolio.png -resize 1200x630^ -gravity Center -extent 1200x630 libreportfolio-og.png

Darktable

As a light room alternative:

Other photo editing

1. GIMP
2. digiKam
3. Photoflow

• Darktable is an open source photography workflow application and raw developer - https://github.com/darktable-org/darktable
• GIMP: The GNU Image Manipulation Program - https://gitlab.gnome.org/GNOME/gimp

Conclusions

Color Grading

A Color Grading LUT (Lookup Table) is essentially a preset color filter that you can apply to your video footage or images to quickly alter and enhance its colors. Think of it as a shortcut for complex color adjustments.

Here’s a breakdown of what that means:

• Lookup Table: At its core, a LUT is a file containing a series of numerical values. These values act as a map, instructing your editing software how to transform the existing colors in your footage to a new set of colors.
• Color Transformation: When you apply a LUT, the software analyzes the original color values of each pixel in your video or image and then uses the LUT’s data to change those values according to a predetermined formula. This can affect various aspects of color, including:
  • Hue: The actual color (e.g., red, blue, green).
  • Saturation: The intensity or purity of a color.
  • Luminance: The brightness or darkness of a color.
  • Contrast: The difference between the light and dark areas.
  • Gamma: The overall tonal range and distribution of brightness.
• Predetermined Look: LUTs are designed to achieve a specific visual style or look. This could be anything from a subtle color correction to a dramatic cinematic effect, or even emulating the look of specific film stocks.
• Efficiency: Instead of manually tweaking numerous color parameters for each shot, applying a LUT provides a one-click solution to achieve a consistent look across your project, saving significant time in the color grading process.

Think of it like this: Imagine you have a recipe (the LUT) that tells you exactly how much of each color ingredient (the original footage colors) you need to mix to get a specific final color dish (the color-graded footage).

Key things to understand about LUTs:

• Not a Replacement for Color Correction: While LUTs handle color grading (the creative aspect of establishing a look), they are generally not a substitute for basic color correction. It’s often best to first balance your footage for white balance, exposure, and contrast before applying a creative LUT.
• Different Types of LUTs:
  • Technical LUTs (Conversion LUTs): These are designed to convert footage from a specific camera’s log format (a flat, high dynamic range format) to a standard color space like Rec. 709, providing a more natural starting point for grading.
  • Creative LUTs (Look LUTs): These are designed to impart a specific aesthetic or style to your footage, such as a vintage feel, a cinematic look, or a specific color palette.
• File Formats: Common LUT file formats include .cube, .3dl, and .look.
• Software Compatibility: LUTs are widely supported by most professional video editing and color grading software (e.g., DaVinci Resolve, Adobe Premiere Pro, Final Cut Pro).

In essence, a color grading LUT is a powerful tool that offers a quick and efficient way to apply complex color transformations to your footage, helping you achieve a desired visual style and maintain consistency across your projects.

• Applying LUT with KDEnlive + .Cube’s for the DJI Oa5pro: https://www.dji.com/pl/downloads/softwares/dji-osmo-action-5-pro-d-log-m-to-rec-709-vivid-lut

FFmpeg + LUT

ffmpeg -i input.mp4 -vf lut3d="path/to/your/lutfile.cube" output.mp4

ffmpeg -i input.mp4 -i haldclut.png -filter_complex haldclut output.mp4

Yes, you can definitely apply .cube files to any MP4 video file using the right software. Here’s how it works with the tools you mentioned:

1. Kdenlive:

• Yes, Kdenlive has a built-in “Apply LUT” effect that directly supports .cube files.
• You can apply this effect to individual clips, tracks, or even the master track to affect the entire project.
• You’ll need to import the .cube file through the effect’s settings.

2. FFmpeg:

• Yes, FFmpeg can apply .cube LUT files to MP4 videos using the lut3d video filter.
• You’ll need to use the command-line interface and specify the path to your .cube file in the filter options.

In general, .cube is a widely supported LUT format in many video editing and color grading applications beyond just Kdenlive and FFmpeg. This includes popular software like:

• DaVinci Resolve
• Adobe Premiere Pro
• Adobe After Effects
• Final Cut Pro

The process generally involves:

1. Importing the MP4 video into your chosen software.
2. Applying a LUT effect or filter to the video clip.
3. Loading or selecting the .cube file within the LUT effect’s settings.
4. Rendering or exporting the video to permanently apply the color grading.

So, regardless of the MP4 file itself (as long as it’s a standard, decodable video), you can use .cube LUTs to change its color appearance using compatible software like Kdenlive or FFmpeg.

D-log

Shutter Speed

Yes, in most standard video production scenarios, it absolutely makes sense and is generally recommended to have your shutter speed roughly double your frame rate. This is often referred to as the “180-degree shutter rule.”

Here’s why it’s a good practice:

• Natural Motion Blur: This combination creates a level of motion blur that looks natural to the human eye. It mimics how we perceive motion in the real world. Each frame captures movement over roughly half of the time it takes for the next frame to be captured.
• Cinematic Look: This is a key element in achieving a cinematic feel. Too little motion blur can make movement look stroboscopic and unnatural, while too much can make everything appear smeary and soft.
• Standard Practice: It’s a widely accepted and followed guideline in filmmaking and video production, making your footage fit seamlessly with other professionally shot material.
• Balance: It strikes a good balance between sharpness and motion portrayal. Objects in motion will have a natural blur, helping to convey speed and movement without sacrificing too much detail in static parts of the frame.

Here’s how it breaks down with common frame rates:

• 24 fps (cinematic standard): Shutter speed around 1/48th of a second (often rounded to 1/50th on many cameras).
• 30 fps (common for broadcast and online content): Shutter speed around 1/60th of a second.
• 60 fps (for slow-motion playback at 30 fps): Shutter speed around 1/120th of a second.
• 120 fps (for even slower slow-motion at 30 or 60 fps): Shutter speed around 1/240th of a second.

When might you deviate from the 180-degree shutter rule?

While the double frame rate rule is a great starting point, there are creative reasons to deviate:

• Faster Shutter Speed (less motion blur):
  • Stroboscopic Effect: To create a sharp, almost stop-motion-like effect, often used for action sequences or dreamlike states (e.g., the opening of “Saving Private Ryan”).
  • Increased Clarity in Fast Motion: To reduce motion blur and make fast-moving objects appear sharper.
• Slower Shutter Speed (more motion blur):
  • Dreamy or Ethereal Look: To create a soft, flowing feel.
  • Exaggerated Sense of Speed: To make fast-moving objects leave long trails of blur.
  • Low Light Situations (as a last resort): If you absolutely need more light and can’t open your aperture or increase ISO further, a slower shutter speed will let in more light, but at the cost of increased motion blur.

In conclusion:

For the vast majority of your video work, aiming for a shutter speed that is double your frame rate is a solid foundation for achieving natural-looking and aesthetically pleasing motion. Deviating from this rule should be a conscious creative choice made to achieve a specific visual effect.

F/OSS Tools for Video

• Create Animations with Code: https://github.com/motion-canvas/motion-canvas
  • https://github.com/motion-canvas/motion-canvas?tab=MIT-1-ov-file#readme

A TypeScript library for creating animated videos using the Canvas API.

Visualize Your Ideas With Code