GlideVideo Studio — User Guide & FAQ

Thirteen local video tools in one Windows desktop app. Trim, clean, repair, assemble, caption, upscale, and convert — your files never leave your machine.

Getting started

GlideVideo Studio is a Windows-only desktop app available on itch.io and the Microsoft Store. Everything runs locally on your PC.

Requirements

Your first job in four steps

  1. Import. Drop files onto Home, the Media page, or any tool's input slot — or press Ctrl+I. Clips land in the Media Bin (the strip along the bottom) with a preview.
  2. Send. Right-click any clip → Send to → pick a tool. You can also drag a clip onto a tool in the left rail, or open a tool and drop the clip into its input slot.
  3. Run. Pick your settings and hit Run. The job appears in the Jobs bar with live progress; the result is added to the Media Bin automatically when done.
  4. Chain. Send the output straight to the next tool (the completion toast suggests likely next steps), or build a reusable Pipeline.
Supported import formats: video (mp4 mov mkv avi webm m4v ts mts wmv), audio (mp3 wav m4a aac flac ogg), and images (jpg png webp bmp tiff) for slideshows.

Core concepts

Your originals are never touched

Every tool writes a new file to your output folder (default Documents\GlideVideo Studio\Output, changeable in Settings). Removing a clip from the Media Bin never deletes anything from disk.

Artifacts, lineage, and generations

Each output is tracked as an artifact that remembers which files it came from and its generation — a counter that goes up by one every time a file passes through a lossy re-encode. A generation-4 file has been re-compressed four times and will show it. If your generation count is climbing, go back to the original and redo the chain, or use Convert's Editing master (CRF 14) preset between steps for a nearly transparent intermediate.

Lossless vs. re-encode

Operations that only rewrap or cut streams (Remux to MP4, Swap Audio with "copy" audio, Join's lossless attempt) are marked lossless — instant and quality-perfect. Anything that changes pixels (resize, transitions, captions, filters) re-encodes with sensible H.264 defaults (typically CRF 18–23).

The job queue

All tools share one queue that runs one heavy job at a time — queue as many as you like and they'll run in order. Cancel any job from the Jobs bar or Jobs page. Failed jobs show the FFmpeg error tail so you can see what actually went wrong.

The 13 tools

Prep & Clean

✂GlidePrep

Trim rough starts and ends, and reframe to a target format (16:9, 9:16, 1:1, 4:5). Three fill modes when the aspect changes: Crop (with position anchors), Blur fill (blurred copy behind), or Matte (black bars). A visual timeline shows exactly what's trimmed vs. kept, with optional scene-boundary snapping.

◌GlideErase

Remove embedded logos, watermarks, and marks — entirely on your machine using a local diffusion inpaint. Draw one or more areas (drag to move, 8 resize handles), set padding and edge feather, and pick a quality. Supports moving watermarks: give each area its own time ranges. Renders at playback speed; output is MP4 or WebM.

♫GlideAudio

Clean up speech: six 0–100 sliders (Noise, Clarity, De-echo, De-hum, Leveling, Limiter), named presets (Clean Voice, Noisy Room, Screen Recording Voiceover, Podcast Speech, Social Clip Speech, Loudness Match Only), loudness targets including "Preserve source loudness," and WAV/MP3/AAC or video output.

▸GlidePace

Find pauses, filler words, and repeated starts using transcription, then review each proposed cut — accept or reject before anything is applied. Or use the quicker silence-only tighten. Great for talking-head recordings and voiceovers.

✚GlideRecover

Repair damaged or interrupted-transfer files (broken index, truncated download, camera crash). Accepts multiple files and repairs each as its own job, with a per-file status list. Repairs are reported conservatively — outputs are not marked lossless.

Assemble

⧉GlideJoin

Merge clips in a fixed order. Auto strategy first tries an instant lossless merge (works when all clips share the same codec, resolution, and framerate), and automatically falls back to a normalizing re-encode when they don't. Or force Re-encode with draft / balanced / clean quality.

◐GlideBlend

Two ways to bridge exactly two clips: Smart cut analyzes the last two seconds of A against the first two of B, finds the closest-matching frame pair, and cuts there for a nearly invisible junction; or pick one of 12+ transitions (fade, dissolve, wipes, slides, circle, radial, blur) with a custom duration and matching audio crossfade.

∞GlideLooper

Turn a short clip into a seamless loop using a seam-swap: pick a seam point (the slider seeks the preview live), and the tool plays the tail first, head second, hiding the join with a blended transition inside the clip. Repeat up to 10× for longer loops. A one-click draft preview lets you check the seam before the full render.

⇋Swap Audio

Replace a video's soundtrack, add an extra audio track, or strip audio entirely (mute) — all without re-encoding the video stream. Audio is stream-copied when the container allows it, with an AAC re-encode option for WAV/FLAC sources going into MP4. Optional "shortest" trims to the shorter of the two inputs.

Publish

▯GlideShorts

Analyzes a long video (audio energy + visual change) and ranks candidate moments for short-form clips, with score bars. Click a candidate to preview it, then export to 9:16, 1:1, or 16:9 with blur / crop / matte framing. Checkboxes + "Export selected" batch several candidates at once.

ＴGlideCaps

Generate captions from speech (local Whisper by default, AssemblyAI optional), or import SRT/VTT. Edit every line, add lines at the playhead, then burn in with full styling: five style presets (Clean, Bold Social, Creator Mono, Lower Third, Highlight Word), font family, size, bold/italic, colors, outline width, background box, and top/center/bottom position.

2xGlideUpscale

Increase resolution 2×, 3×, or 4×. AI mode uses Real-ESRGAN (three models incl. photo and anime variants) when a Vulkan GPU is available; otherwise a fast, high-quality FFmpeg Lanczos scale runs instead. Audio is carried over untouched where possible.

◇GlideUtilities

The Swiss-army knife: resize, crop, rotate/flip, reverse, cut a time range, brightness/contrast/saturation filters, exact speed change (0.25×–4× with pitch-corrected audio), freeze last frame, frame interpolation to higher fps, aspect fit/fill, JPG/PNG snapshots and numbered frame sequences, quick MP4/WebM convert, video→GIF, compress-to-a-max-file-size, and images→video slideshow.

⇄GlideConvert

15 plain-language presets in four groups — Delivery (Android, YouTube, High quality, Standard, Editing master, Small file), Social (Instagram square, TikTok vertical, WhatsApp), Remux & web (lossless MP4 remux, WebM VP9), and Extract audio (MP3, AAC, WAV, FLAC). Drop several files in to batch-convert them. An Advanced mode exposes container, codec, CRF, preset, audio, scaling, fps, and raw extra FFmpeg args — and you can save your knob combinations as named custom presets.

Quick utility map — "which tool do I use?"

TaskUse
Resize / crop / rotate / flip / reverseUtilities
Cut a time rangeUtilities (quick cut) or Prep (trim with visual timeline)
Change speedUtilities for an exact speed; Pace for tightening silences
Merge clipsJoin (fixed order); Blend for a transition between two
Freeze last frame / interpolate framesUtilities
Change aspect ratioUtilities (fit/fill) or Prep (crop/blur/matte with anchors)
Save a JPG/PNG frame or frame sequenceUtilities
Convert to MP4/WebMUtilities for one quick file; Convert for presets and batches
Video → GIFUtilities (palette-optimized, start/length/fps/width controls)
Fit under a size limit (e.g. 8 MB)Utilities → Compress (set max MB; bitrate is computed for you)
Photos → slideshow videoUtilities → Images to video
Remove a watermarkErase
Fix a broken fileRecover

Pipelines — chain tools automatically

The Pipelines page is a visual, linear node builder. Pick tools from the palette (everything that can run unattended: Recover, Prep, Pace silence-tighten, Audio, Looper, all Utilities operations including GIF and Compress, Upscale, Caps auto-burn, Convert), order them vertically, tune each node's settings in the inspector, then Run the whole chain on a clip. Each step's output feeds the next automatically, and every intermediate still lands in your Media Bin with full lineage.

Projects, presets, and session restore

Settings

SettingWhat it does
FFmpeg / FFprobe pathLeave empty to auto-detect on PATH (or a bundled bin\ folder). Set explicit paths if detection fails — any existing FFmpeg install works.
Real-ESRGAN pathOptional. Point at realesrgan-ncnn-vulkan.exe (or its folder) to enable AI upscaling. The packaged app ships its own copy; Upscale falls back to FFmpeg scaling when AI isn't available.
Output folderWhere all results are written. Default: Documents\GlideVideo Studio\Output.
Transcription providerLocal Whisper (default) keeps everything on your machine — the model downloads once on first use. AssemblyAI (cloud) is optional for higher accuracy; paste your API key. Used by Caps and Pace "Find cuts."

Keyboard shortcuts

KeysAction
Ctrl+1…9, 0, -Switch tools in rail order
Ctrl+IImport media
Ctrl+clickMulti-select clips in the Media Bin / Media page (for batch Send-to and batch convert)
Mouse wheel over Media BinScroll the bin horizontally

FAQ

General

Does GlideVideo Studio upload my videos anywhere?
No. All processing — including watermark removal, transcription, and AI upscaling — runs on your PC. The single exception is optional AssemblyAI cloud transcription, which only happens if you explicitly select it in Settings and add your own API key.
Will the app modify or delete my original files?
Never. Every tool writes a new file to your output folder. "Remove from bin" only removes the entry from the app's working set — nothing on disk is touched.
Where do my outputs go?
Documents\GlideVideo Studio\Output by default (change it in Settings). Finished clips also appear in the Media Bin with "Reveal" to jump to the file in Explorer.
Is there a Mac or Linux version?
No — GlideVideo Studio is Windows-only.
Do I need a license key?
No. Purchase is handled entirely by the storefront (itch.io or the Microsoft Store). There are no keys, accounts, or feature gates.

Quality & formats

What does the "lossless" tag on an output mean?
The video/audio data was copied bit-for-bit without re-encoding (e.g. Remux to MP4, Swap Audio with copy, Join's lossless merge). These jobs are near-instant and lose zero quality.
What is the "generation" number on a clip?
How many lossy re-encodes the file has been through since its original. Each re-encode stacks a little quality loss. Keep chains short, or use Convert's Editing master (CRF 14) preset as a high-quality intermediate when you must chain many steps.
Why did Join re-encode my clips instead of merging losslessly?
The lossless merge requires every clip to share the same codec, resolution, framerate, and compatible timestamps. When they don't, the auto strategy falls back to normalizing everything to a common size/fps and re-encoding — that's expected, not an error.
My GIF has no sound.
GIF as a format has no audio — that's true everywhere, not just here. If you want a short silent-looking clip with sound, use Utilities → Convert to MP4/WebM instead.
Compress: how does the size target work?
You set a max size in MB. If the file is already under the cap it's losslessly remuxed. Otherwise the tool computes the video bitrate from your clip's duration and audio budget (reserving ~6% for container overhead) and encodes; if the result is still over, it retries at lower bitrates and smaller resolutions automatically. If even the smallest attempt can't fit, you'll get a clear error suggesting a larger cap, muting audio, or a shorter cut.
Which speed range does Utilities support?
0.25× to 4×, with pitch-corrected audio (chained atempo). For slow motion beyond that, consider Interpolate first to raise the frame rate.

Captions & transcription

Do captions require the internet?
No. The default is a local Whisper model (base.en) that downloads once on first use, then works fully offline. AssemblyAI is an optional cloud alternative.
Transcription seems instant the second time — why?
Transcripts are cached per file (keyed to the file's path, size, and modification time). Caps, Pace, and pipelines reuse the cache instead of re-transcribing.
Can I bring my own subtitles?
Yes — Caps imports SRT and VTT files, which you can then re-style and edit line-by-line before burning in.
Are burned-in captions removable later?
No — burn-in renders text into the pixels permanently. Keep your pre-caption file (it stays in your bin and on disk) if you might need a caption-free version.

Upscaling & performance

Why is AI upscaling greyed out / falling back to FFmpeg?
AI mode needs the Real-ESRGAN Vulkan binary and a working Vulkan GPU. The app probes your GPU at startup; if the probe fails (very old GPU, missing drivers, some VMs), Upscale uses high-quality Lanczos scaling instead. Updating your GPU drivers usually fixes Vulkan detection.
AI upscaling is slow — is that normal?
Yes. AI mode extracts every frame as PNG, upscales each through the neural net, and reassembles the video. A few minutes of footage can take a long time depending on GPU and resolution. The FFmpeg mode is dramatically faster when you just need more pixels.
Why does only one job run at a time?
Video encoding saturates your CPU/GPU; running jobs serially is almost always faster in total and keeps the machine usable. Queue freely — everything runs in order.

Erase (watermark removal)

How long does Erase take?
Erase renders in real time — a 3-minute clip takes about 3 minutes. It processes the video frame-by-frame in the app itself (not FFmpeg), which is what allows fully local inpainting.
The watermark moves around during the video.
Use the moving-watermark mode: create an area per position and give each its own time ranges ("Start here" / "End here" at the playhead). Only the active ranges are inpainted at any moment.
What output format does Erase produce?
MP4 where supported, otherwise WebM (it depends on your system's media encoder support). You can always run the result through Convert afterwards.

Troubleshooting

"FFmpeg not found — set its path in Settings."
FFmpeg isn't on your PATH. Install FFmpeg and either add it to PATH or set the full paths to ffmpeg.exe and ffprobe.exe in Settings → FFmpeg, then check the "Detected" status line turns green.
A job failed — where do I see why?
Open the Jobs page. Failed jobs keep the tail of the FFmpeg error output. Common causes: an unsupported audio codec for the target container (try the AAC option where offered), a corrupt source (run it through Recover first), or trying to stream-copy while also scaling (in Convert Advanced, "Copy" video codec can't be combined with scaling or fps changes — pick a real codec or set Scale to "Keep source size").
Swap Audio failed with my WebM/MKV file.
The output keeps your video's container, and not every audio codec fits every container (e.g. MP3/AAC into WebM). Either remux the video to MP4 first (Convert → Remux to MP4) or choose the AAC re-encode option with an MP4/MOV/MKV source.
Transcription fails or hangs on first use.
First local transcription downloads the Whisper model (needs internet once). If it fails outright, check the Jobs error text; as a workaround you can switch to AssemblyAI in Settings, and switch back later.
The preview won't play but the file is fine elsewhere.
In-app previews use the system media stack, which can't play every codec (e.g. some 10-bit or exotic formats). The tools will still process the file correctly — or remux/convert to H.264 MP4 for a previewable working copy.
My output looks softer than the original.
Check the generation count. Each lossy step re-encodes; three or four stacked steps show visible loss. Redo the chain from the original as a single pipeline, or insert Convert's Editing master (CRF 14) preset between heavy steps.
Scene-snap / Shorts analysis takes a while on the first run.
Scene detection scans the whole file once, then results are cached per file — subsequent runs of Prep snapping, Blend, or Shorts on the same clip are much faster.