Real project · Desktop app · Content
A desktop tool that batch-converts horizontal 16:9 videos into vertical 9:16 for TikTok, Reels and Shorts: centre crop, blurred background or black bars, a caption from the file name and fast GPU encoding.
Owners of clip channels convert dozens of horizontal videos to vertical every day. In a video editor that is pure routine: every file on its own, the same actions, the same caption, minutes of export per clip.
They needed a tool they could drop a whole folder into and get finished vertical videos with captions, without opening an editor.
The app takes files or a folder by drag and drop, converts everything in one run in the chosen mode, adds a caption from a file-name template and encodes on the GPU through NVENC.
Results go into a 9x16_converted subfolder next to the originals, metadata is stripped and every step is visible in the log.
Drag clips or a whole folder into the window. The output folder can stay next to the originals or be set anywhere.
Crop cuts out the centre of the frame, Blur puts a blurred copy of the video behind it, Black adds black bars. "All three" produces three versions from one clip.
The {filename} template inserts the file name, supports line breaks, and the caption can be edited per file. Font, size (48, 72 or 110 px), position at the top, centre or bottom, optional upper case.
Mirroring and light noise when needed, so the clips differ from the original.
The app checks whether h264_nvenc is available and encodes on the GPU at maximum quality (CRF 18). Without an NVIDIA card it falls back to software encoding.
Finished files in the subfolder, metadata removed, and every operation and error in the log.
A whole folder in one run, with progress and a log for every file.
Crop, Blur and Black together: three versions of a clip to test which format performs better.
The caption list is edited per video; "reset all" restores the template.
Any font installed on the system, with accurate text wrapping through Pillow.
NVENC makes export several times faster than the CPU, with an encoder check at start-up.
All metadata from the original is removed from the finished files.
A horizontal clip becomes vertical in one pass, with no video editor.
FormatFraming modes: crop, blurred background, black bars, or all three at once.
ModesRun for a whole folder: files are processed one after another with no one watching.
RunMaximum encoding quality on the GPU through h264_nvenc.
QualityPython for the interface and the file queue, FFmpeg for framing and encoding, NVENC for speed on NVIDIA cards, Pillow for rendering captions with correct line wrapping.
Have an idea?
