Hi this is an experimental project idea, i have a few months to explore. Im giving a keynote by Zoom remotely from UK to Paris in the summer approximately one hour, two speakers, host in Paris, - if i screen recorded the view on one Mac, saved it, within the 90 minute window we have to present , is there some kind of random frames grabber that could run through it so it compiles upto 4 minutes of nonsensical cuts, without understanding our words, and remix the joint lecture into a new and possibly provocatively interesting to artists version, that i would then stream back from my zoom presenter view to the worldwide audience ? If not VDMX, does anyone imagine TouchDesigner being capable, I have time to learn. many thanks for considering the challenge, file length, saving, compression, conversion and then random remixer - script ? FX ? i dont want to apply glitch or consciously select content from the talk … it must be random, its connected to Burroughs
You could do it with either TouchDesigner or VDMX. The issue with VDMX, is you’ll be sending the random video clips to a folder and compressing them together after. If you keep them in the same format, the encode times will be low, but you’ll ultimately want to share via h.264 or h.265 over the web.
Might be worth posing to the TouchDesigner community to figure out how to record / pause an H.264 video file randomly (and let it finalize compression at the end or in chunks).
I’ve used VDMX to do similar things in the past. Use a random data source to trigger the Movie Recorder to capture a layer with a Vid In (Windows in other Applications). Then use VDMX to randomly trigger the captured clips to stream or record the recompiled clips. The interesting part here is deciding what the random data source is and how it relates to the Zoom session or not.
thANKS, good to know its possible and can be tested out soon… hoping to get splicer d’cutup’ words from the zoom lecture, possibly a few words in each clip that may create a new text, idea or offshoot
yeah the random data source, maybe from some Burroughs recordings
You could use Adobe Premier Pro (or a browser based service) to transcribe the video from the Zoom session to a text file. Then randomise the playback of the text to create a cut-up style visual in VDMX. To randomise the text you could do this in a text editor like TextWrangler or BBEdit to create a static file or create a custom app with Syphon output or VDMX source in something like Vuo (https://vuo.org/).
do you think this can happen quite fast ? I have a 90 minute window and hoped to capture / cutup the lecture and Q&A from the first 70 minutes… then deliver that file from my presenter window at the end - i have 1gIG ethenet connecting mac to fibre broadband approx 10 / 40 speeds here. thanks for your indeptth idea - i have months to test this… so all options can be considered and tested
If H.264 is your end product, you can try to Merge files instead of exporting them. I’ve done this with .MP4Tools, and MKV tools, in a way that does a pass thru with audio and video and skips the re-encode portion of the video.
But maybe @2bitpunk method is better.
Um, it’s going to be difficult to automate this using Zoom App transcription tools because of the timing. I would suggest manual transcribing the text using an off screen helper (real human). The transcriber could enter the text string / sentence into a VDMX text layer (stylised as the original Burroughs paper cut up). And use the Movie Recorder plugin to capture a PNG. Then you could randomly playback the PNG files combining one or more on VDMX output. This would give you the quickest turnaround but but introduces the problem of curation… what text to transcribe, how quick can you transcribe, etc.
One thing you might try is Google Chrome > Google Docs Voice typing. For a group talk the conversation will be captured without punctuation. But the advantage with Docs Voice typing is that you can edit as it types. For best results you will need to set up your audio routing to target Zoom for the input, something like Loopback would help with this. This method is pretty robust and would get the conversation into raw text for zero cost which is half the battle.
Source (random round table conversation): https://youtu.be/c6C7sm8LcPU?t=57