Launch tomorrow: what to expect

Tomorrow we’re opening Lucid to the public. This has been a fast build—about two and a half weeks—and today’s work was mostly last‑mile clarity: making the website match what the bot actually offers, tightening a couple of diagnostics/quoting edges, and continuing to harden payment operations behind the scenes.

Launch note: going live tomorrow

Lucid has a lot of moving parts: a Telegram-native generation workflow, a credits + crypto payment pipeline, and a website that serves as the canonical landing page and documentation hub.

What’s in place going into launch:

• Generate videos from text prompts or images inside Telegram, with model selection and core output controls.
• A credit system with balance and invoice history, plus automatic refunds when a generation fails.
• On-chain credit purchases with invoices and automated credit assignment, hardened to prevent old payments being misapplied to new invoices.
• Reliability work throughout: clearer error handling, sensible retries, and a growing test suite to catch regressions early.

Launch-day expectations

If you’re trying Lucid for the first time:

• Send /start to see onboarding and the available commands.
• Use /balance to confirm your credits.
• Try a simple text prompt first (text-to-video).
• Then send an image to animate it (with or without a caption).

What “normal” looks like on launch day:

• You may hit short processing queues.
• A generation can fail occasionally (model constraints, network issues). When a job fails, credits are automatically refunded.
• Your past outputs live in /history. Payment/invoice records live in /check_invoices.

If something fails: retry the request, check /balance, and check /check_invoices. /help and the bot’s health check command (/health) are there for diagnostics.

Website: FAQ accuracy pass

We corrected pricing information and model names in the site FAQ so it matches what the bot actually offers.

Website: video gallery interactions

We improved the video browsing experience on the site: video cards are more interactive now, videos open in a modal for easier viewing, and the modal interactions and styling are smoother.

Website: featured videos on the homepage

We added a featured videos section to the homepage, so the site can highlight recent or standout generations without making you dig through the gallery.

Under the hood: treasury automation (transaction processing)

We continued internal treasury automation work, expanding the database layer and tightening transaction processing so payment operations stay predictable as usage grows.

We also did a small cleanup pass, removing deprecated internal tooling/UI pieces that are no longer used.

This didn’t change the Telegram generation workflow.

Bot: health check wording

We updated the health check command output to reflect an API naming change, so diagnostics stay aligned with the system.

Bot: image-to-video model quotes

We improved how the bot fetches quote data for image-to-video models by generating a small placeholder image when one is required.

This is a behind-the-scenes consistency fix: it helps keep model quote information available without changing the generation workflow.

(Implementation note: this also added an image processing dependency to support the placeholder image generation.)

---

User-facing summary: the website is clearer and easier to browse (FAQ accuracy, gallery improvements, and featured videos). On the bot side, diagnostics and model quoting are a bit more consistent. Today’s operations work focused on keeping payments predictable as usage grows.