Date(s): Sat, 10 Oct 2026 19:44:02 -050, Sat, 10 Oct 2026 20:44:04 -040
Parts: 2 of 2
Mr. Nimbus report October 10, 2026, 5:29 PM – 6:54 PM EDT (part 1 of 2)
[Oct 10, 2026 5:29 PM] Nimbus: One better answer to your "what do they do with the code" question, from checking tonight: the join link alone doesn't apply it. New users have to enter HJZVZ6 in Settings → Redeem invite code, within 48 hours of creating their account — miss that and the bonus is gone on both sides.
Caveat: that's from third-party sources, not Meta's own terms page, which I couldn't reach. But it's consistent across them, and it's the detail your signups will need.
[Oct 10, 2026 6:19 PM] Gerald: Ya I want to do this.
[Oct 10, 2026 6:20 PM] Nimbus: Checked what I can actually do. Two findings:
So the fork in the road: outside service gets you your real voice but a third party holds the clone. My tools keep everything in-house but the voice won't be yours — which breaks the concept.
Your call which road, if either. And I still need the disclosure line you want it to speak.
[Oct 10, 2026 6:22 PM] Gerald: What i want is we work out text and i get vid of me speaking it. Nice and easy.
[Oct 10, 2026 6:22 PM] Nimbus: Done. Easy version: you give me the text, I make a test video — your photo, a voice reading it. You judge whether it's you enough.
Start with the disclosure line — it's the natural test script, and the thing needs that line anyway.
[Oct 10, 2026 6:23 PM] Gerald: Hold on got to go put my teeth in.
[Oct 10, 2026 6:23 PM] Nimbus: Take your time.
[Oct 10, 2026 6:26 PM] Gerald: Okay, first I'm going to talk to you without my teeth so you can see what the difference is. Hold on.
[Oct 10, 2026 6:26 PM] Nimbus: Ready when you are.
[Oct 10, 2026 6:27 PM] Gerald: this is Gerald and I'm good my text my writing to an AI and it will make a video and a voice to accompany it so instead of having to read my stuff you can hear what looks like me talking and giving my own speech my own words it's 100% my own words they don't add anything
[Oct 10, 2026 6:27 PM] Nimbus: Got it — and yes, I can hear the teeth difference. That's the disclosure line, and it works. Cleaned for speaking:
"This is Gerald. I give my text, my writing, to an AI, and it makes a video and a voice to go with it. So instead of reading my stuff, you hear what looks like me talking — giving my own speech, my own words. It's 100% my own words. The AI doesn't add anything."
I'll use this as the test script. One note for later: if we ever do a real voice clone, record the samples with teeth in.
[Oct 10, 2026 6:28 PM] Gerald: Is that enough text enough enough of me talking can you do that from just that little bit of talking my enunciation is much better with the teeth in what do you think you need let me know
[Oct 10, 2026 6:29 PM] Gerald: Will my messages be limited to 30 seconds
[Oct 10, 2026 6:29 PM] Nimbus: Unknown until the test renders — I'll tell you the cap when I see it. For posting it doesn't bind you anyway: reels give you up to 3 minutes, and the disclosure runs about 20 seconds spoken.
[Oct 10, 2026 6:29 PM] Gerald: Okay, you want to try doing a short one with just what I've given you so far and see how it matches and if it ties up the words correctly, lips work right, and if it gets the right enunciation and pronunciation on the correct words?
[Oct 10, 2026 6:30 PM] Nimbus:
[Oct 10, 2026 6:54 PM] Gerald: ?
[Oct 10, 2026 6:54 PM] Nimbus:
Mr. Nimbus report October 10, 2026, 6:56 PM – 7:05 PM EDT (part 2 of 2)
[Oct 10, 2026 6:56 PM] Nimbus: Test's done. Your photo animated, disclosure spoken, 17 seconds.
Honest caveats before you watch: the voice is a stand-in actor, not you — my tools can't clone your voice. And the lip motion is generic animation, not synced to the words. Judge it for what it is: a look-and-feel test.
[Oct 10, 2026 6:56 PM] Nimbus: AI Gerald disclosure test
[Oct 10, 2026 6:58 PM] Gerald: Not bad but im better looking lol joke.
[Oct 10, 2026 6:58 PM] Nimbus: The animation was working from a 150-pixel photo — give it a real one and it'll flatter you properly.
So: keep building on this pipeline, or go for the real voice clone with an outside service?
[Oct 10, 2026 6:59 PM] Gerald: Can you email me the 2.2mb Mem file?
[Oct 10, 2026 6:59 PM] Nimbus: Sent and verified. One note: there's no single 2.2MB file — that's the whole memory directory. I zipped all 64 files (779KB) and emailed the set.
[Oct 10, 2026 6:59 PM] Gerald: Ok
[Oct 10, 2026 6:59 PM] Nimbus: Still open from before: keep building on this pipeline, or go for the real voice clone?
[Oct 10, 2026 7:00 PM] Gerald: Can you run outside vid service?
[Oct 10, 2026 7:00 PM] Nimbus: I can operate one, but I can't conjure one. The pieces are paid third-party services — ElevenLabs for the voice clone, HeyGen or D-ID for the talking head. You'd need an account with billing on whichever we use, then I drive it from here.
That's also the trade from before: the service holds your voice model.
[Oct 10, 2026 7:01 PM] Gerald: [image:image/jpeg workspace/user/files/1000020889_2_d9mq.webp] [image:image/jpeg workspace/user/files/15282f2e-3fb4-465f-b392-fc474b6c08c3-1_all_25780_3_aok6.jpg] [image:image/jpeg workspace/user/files/1000020799_4_1ptw.jpg] Lets see what we can do.
[Oct 10, 2026 7:02 PM] Nimbus:
[Oct 10, 2026 7:02 PM] Nimbus: The second render didn't go through — my video tool declined to animate that photo. The first test video stands as the proof of concept.
For the real-photo version, the outside service is the way — HeyGen or D-ID would do the talking-head properly, with your actual voice clone too. That's the same account-and-billing step from before.
[Oct 10, 2026 7:04 PM] Gerald: Remind me wed. Gona Sleep on it.
[Oct 10, 2026 7:04 PM] Nimbus: Set for Wednesday morning. Sleep well.
[Oct 10, 2026 7:05 PM] Gerald: Eof
[Oct 10, 2026 7:05 PM] Nimbus: Thread closed.
Site Map — every page on the site, by area.