MegaSplats: plan a large capture and choose how to process it
A MegaSplat workflow helps you work with a large geotagged photo capture, such as a site or a broad outdoor area. Start by deciding what you want people to explore and how your photographs cover it. A large map footprint alone does not mean every surface was captured.
The MegaSplat workspace offers processing choices for large captures. Some modes depend on account availability. Review the choices actually shown to you and the current estimate before starting paid work.
Choose the right route
An ordinary video or smaller photo set: start in Create. You do not need the MegaSplat workspace just because a recording covers an interesting place.
A large geotagged photo set, processed in tiles: use the MegaSplat Tiles route. Plan the area and sections, review the proposed work, and follow the project through tile processing and assembly. This route uses the photos; video clips are not its input.
Geotagged photos plus supported drone-video detail: where One solve is available, combine the photographs and selected video frames into one model. Read the photos and video walkthrough. This route needs geotagged still photographs as well as the supported clips; it is not the video-only upload route.
Extra detail in a finished large capture: check Sharpen. That is a separate eligibility check and paid upgrade. It is not the same as choosing tiles before processing or simply loading more detail in the viewer.
Prepare the capture
- Preserve your original photographs, video clips and location sidecars. Use one coherent capture area and check the location metadata before planning work.
- Check image sharpness and overlap. Include different viewpoints of important surfaces, and connect passes with shared features. A wide overview does not replace close coverage of a facade, recess or interior.
- Open the workspace and inspect how the input is located. A map placement is useful context, not proof that the reconstruction will be complete or survey-accurate.
- Choose the available processing route. Review exclusions, size limits and the price before confirming.
- Follow the project status. Uploaded, trained, assembled and ready to stream describe different stages; a completed tile does not establish that the whole scene is ready.
For capture technique, use coverage first, composition later. For placing a separate detailed capture inside a larger overview, use combine and align captures. That is a scene-editing workflow, with its own alignment checks.
Video and SRT location data
An SRT file is often a subtitle file. Only supported location-bearing telemetry is useful here. Ordinary speech captions do not supply camera positions, and not every DJI model or firmware writes a format Twinbly can read.
For supported drone recordings, keep the matching .SRT beside its original video. Use the same filename stem—for example, flight-01.MP4 with flight-01.SRT—and include a matching sidecar for every clip whose location you want to use. Check the upload's notices; the timing must match the video.
There are two distinct routes:
- Ordinary Create: add the video and matching supported DJI SRT files together. Video processing can still be available without readable GPS telemetry; missing or invalid location data does not by itself make ordinary video unusable. Check whether location was accepted before relying on placement.
- MegaSplat One solve: start with at least 80 geotagged still photographs. Eligible video clips need supported matching SRT data so selected frames can join the located photo set. A clip can be excluded if its positions are missing, its telemetry format is unsupported or the browser cannot decode it. The remaining accepted inputs can still be used.
The photos-plus-video route currently allows up to 4,000 total images and roughly a 12 GB upload per job. The photographs are retained in the plan; video sampling may be reduced to fit. These are limits for that route, not a promise that every video or every Twinbly workflow has the same budget. DJI Mavic 3 Enterprise subtitle telemetry is supported; inspect the acceptance notices for your own files.
GPS placement and Sharpen eligibility are separate. If a One solve result is eligible, you can request Sharpen later, with a separate quote. Uploading an SRT file does not add usable image coverage or make an otherwise ineligible scene refinable.
Check the result before sharing
Open the completed scene and inspect the areas and viewpoints that matter to a visitor. Look at transitions between sections, fine details, thin objects and places the camera could not see well. For an assembled project, check its status for incomplete or failed sections before treating it as finished.
Treat ordinary drone GPS as a location aid, not a survey-grade accuracy guarantee. Inspect alignment and the reported accuracy information before making measurement claims.
Then straighten and prepare the scene, set a tour or publish it. You can also combine an overview with separate detailed captures; confirm their alignment and viewing behavior before presenting them as one experience.
Can I keep adding more video detail later?
Keep your originals, but do not assume an existing job will absorb them. Additional footage may need a new processing run or a separate capture that you align into the scene. The current web Sharpen action works from the finished capture's retained images and reconstruction; it does not automatically fetch unused video frames.
If the main issue is a missing side or an unseen space, capture that coverage first. More processing cannot substitute for a view you never recorded.