Stand alone desktop app + more
Z
Zak Melms
Hello - I have a huge library of files. It's over 200TB. It would be fantastic if I could have a desktop version of Jumper for many reasons, but a few scenarios I can think of are:
- Pre-processing all of my footage in the background. I'd love to just set the app loose on my library and have it generate a sidecar file for every video clip and image I have. I'd rather not have to process the media EVERY TIME I START A NEW PROJECT - that's just a headache. Also my desktop is way more powerful than my laptop and that way I can have the office computer do the hard part while the travel unit does the easy bits.
- Using the pre-processing in my searches in Premiere. When I import media into premiere I want Jumper to use those sidecars and not have to process anything.
- Searching on the desktop - I want to find files in my library with the Jumper search functions. Sometimes I don't even know where to start so this would be literally a KILLER APP.
- Generating file exports - I'd like to be able to use the search function on the desktop, create a list of files to use and then either copy or move those files to a new folder/drive/etc. OR use that list to open up a new premiere project and start an edit.
- Adding my own keywords to specific images or items - such as location, subject, etc, that are then searchable in the future.
- Organizing the actual files.
That's just a starting point, but those features would be AWESOME!!
Thanks!
Chris Hocking
FWIW, I personally think a really good solution would be to have a Jumper menubar icon on macOS (i.e. put the Jumper icon next to the clock in the top right corner of your screen) that when you click on it pops out a search panel (similar to Dropbox's menubar icon in design for example), so you can search through all your already analysed footage - but also easily add folders to be analysed. The menubar app can also have a progress bar, showing what's currently processing in Jumper (whether it's been triggered from an NLE or from the menubar). You could then assign a global shortcut key, so you can press a key combination and access the menubar interface - basically similar to how Spotlight works, for instant searching.
M
Max
Zak Melms Check out these new updates coming in the next release, should be of some use to you I think? Posted two short videos on Discord:
M
Max
Hi Zak Melms, thank you for the detailed requests! We're working on some stuff related to this right now but I'll try to go through your points and see if there is something that could be useful in what you can currently do today.
- I would just do an empty new Premiere project and literally dump your entire drive/folder with media in there, and just let it process over night on your desktop. This will populate the Analysis folder you have specified in the Jumper Settings tab.
Then, you can copy this cache directory over to your laptop, and all the footage will be searchable (Jumper doesn't care about the path of files, only their content, so if you have a file with the same content processed on different machines, it will still find it)
- If I understood correctly, this would be achieved by doing as explained in 1., then just setting the Analysis folder to be the one you used when you processed everythying (or where you copied that folder to the other computer)
- This is being actively worked on, so that you would have a toggle that says "search in all processed footage", which would search everything regardless of if it is imported in a project or not. Then we plan to add a button to locate the file on disk if it is not in your project, so you can import it intop your project.
- Bulk actions - got it. We're currently working on an instruction-based solution for creating a rough cut/timeline of selects. But this is not quite what you're asking for. We could combine functionality of 3. with some bulk action to import perhaps? Like "import all clips from current search results" after you've searched using the "search in all processed footage" option.
- Since this is already possible in your NLE (keywords in Premiere, tags in FCP etc), I'm not sure it's in scope for what we want to achieve. One thing that is related to 4. (rough cut/timeline of selects) is that we have the ability to also generate a list of keywords for each "scene" Jumper finds in the analysis. Perhaps that would be useful?
- Could you elaborate? Organizing files in filesystem or in your NLE is obviously something you can do manually, but I understand you want Jumper to help out to do this in some way? What's the ideal scenario?
M
Max
Regarding 1., ff it's 200TB of footage it may be a bit much for Premiere to handle if you drop all of it at once. Perhaps do it in batches. But you in either case you should only ever need to process footage once if you always use the same Analysis folder/cache directory, and use the same model. If you're finding that you have to re-process some file that you are sure you already processed using the currently selected visual search model, contact me by DM on Discord or send an email to hello@getjumper.io.
Z
Zak Melms
Max
thanks for the reply, it's great to hear this is very much an active bit of software and that gains are being made all the time. I just updated today to the latest. Excited to see how the changes are working.
regarding #5 - the magic bullet I'm looking for is adding keywords to media in either a sidecar file, metadata, or something else. To the best of my knowledge if you keyword inside Premiere those keywords won't travel with the media outside of that one project. When dealing with a Library of footage, I'd love to be able to identify clips by location, which is often how I organize them, OR tell JUMPER that THIS CLIP is showing a particular location or specific subject matter. Basically, I wonder if it's possible to let users add some more data to the model so that it gets smarter and learns more about my footage library. Like specific people so you can say find TOM, or show me all of the footage from this location, or "this image has someone wearing a hardhat and safety vest. I might just need to do a few more projects now that the entire cache is searchable
6 - I'll circle back to this :)
M
Max
Zak Melms Gotcha. For now, I would use Jumper with the new feature to "Search all analysed media" to locate the clips you want, then do something to organize the media yourself such as adding Premiere keywords or putting them in a folder or something. The thing about telling Jumper that "this clip shows a particular location or subject matter", is that's what we're already giving you the ability to locate by just typing in that search. It feels a bit like we're going in circles if you 1. use jumper to find X, then 2. tell Jumper "this is X" - Jumper already knew it was X since you could just search for it and locate it. But face detection like "this is TOM" is on the TODO list as I think I mentioned, so that will be possible in a future release.
Z
Zak Melms
Max - I'm going to re-bump the desktop app :) as this feature is critical. Especially if Jumper can access a universal cache of files, right now, I have close to 10,000 clips and photos analyzed, and I have 5x that to go. Here's my reason - Premiere chokes when it tries to load all of these clips, or even a small section of them, to do "batch Jumper processing". With a desktop functionality, I just want to aim the app at a folder and say - "analyze all of this so it's searchable". And call that the "CLIENT 1 Analysis Cache" without having the layer of Premiere running on top of that function.
Then, I don't need to load all 100,000 photos and videos into Premiere to make them searchable, which does not work. I just used the cached library.
The second reason is that as the algorithm improves, I don't want to go back to all these older Premiere projects to "re-analyze" the footage. I'd like to just go to the Jumper APP and say - re-analyze everything in this particular cache (or media folder). Then, let the computer chug overnight or for a few days. Opening up dozens of projects and hoping that Premiere doesn't crash or load fast enough is not really functional.
m
mgl
Zak Melms With the new feature to "search in all analysed footage in current analysis folder" you can have a Premiere project with no files at all and just hit that switch to search in all your files anyway. Premiere would just be "GUI", you wouldn't have to import anything. You could just change the folder to e.g. "CLIENT 1 Analysis Cache" to search that cache.
Z
Zak Melms
mgl I hear you and that works well. I'm already utilizing that feature. What's missing is the ability to process media outside of Premiere because the process of loading all of my media into a Premiere project is tedious and could lead to an incomplete scan of the media. The Jumper "APP" already exists I just want to add functionality to it something like "pre-analyze". Where I can point it to a folder and have it analyze everything inside that folder without opening Premiere.
And as I said, as the algo improves and the cache files shrink I want to say - "re-analyze" and have Jumper go find all the clips from the cache and re-analyze them on it's own, dumping the older or larger cache files in favor of more efficient ones.
M
Max
Zak Melms One thing we could do is to add another "process mode" where you just enter/locate a path to a directory (or several directories) and we process each valid media file we find there (perhaps with some user-defined filters to e.g. ignore certain file extensions), without ever importing anything into Premiere (or the NLE from which you run Jumper). This would still require you to run Jumper from the NLE, but it would be just like a "frontend" - instead of running it via e.g. Chrome or Safari (like it is possible now, just not released publicly).
With regards to your thoughts on "re-processing" - this is already supported when you switch visual search models under Settings. Although we don't currently delete the cache files in order to save storage because we want to maintain backward compatibility. But it might be a good idea to offer a simple "optimise cache size" button somewhere, where we go in and replace the old large cache files with the new format, shrinking the storage size by about 90%.