Glycol reconcentrators overlay water content on silent Wan 2.5 I2V, 15cr

Photograph the reconcentrator you ship; Wan Video 2.5 I2V is audio-false, 1080p, 5s or 10s, 15cr so the firetube does not morph — water content is overlay; the T2V sibling at the same 15cr carries native audio if you actually want a bed.

A glycol reconcentrator clip that books is the firetube and the still column on this skid. Tube diameter, burner train, weir, the lean-glycol outlet you actually flange — those are the SKU. Water content is a number from a sample or a spec sheet. Motion belongs on a photograph of the unit. Type belongs after picture lock. A paragraph that rebuilds a TEG reboiler in a field that is not yours is a mis-ship.

Photograph the reconcentrator you ship

Even light. Firetube large in frame. Stack, sight glass, and burner plumbing readable if they ship on the package. That file is the print. Text-to-video will invent a cousin reboiler with a different tube and a cleaner paint job. The buyer is matching iron, not atmosphere.

The AI image-to-video tool is the door. No still, no generate on this job. Frame one is the file. Re-describing the firetube in the prompt is how the model starts drawing a second tube.

Prompt camera and physics only. A slow hold. A light rake across the shell. Camera locked. No new nozzles. No new skid steel. If the still includes a nameplate or a serial, keep it. Do not prompt it off and hope the model rebuilds the reconcentrator.

Manufacturers is the plant page. This leftover is one reconcentrator, one firetube. Do not rewrite the plant tour as a talking dew-point lesson.

If the still would fail a cut-sheet audit — wrong burner, extra exchanger — every second of video is a mis-ship. Shoot again.

Silent Wan 2.5 I2V holds the firetube

Wan Video 2.5 Image to Video is high-quality image-to-video. It requires an image. Audio is false. Durations: 5s and 10s. Ceiling 1080p. Display price 15 credits. Quality chips: 480p, 720p, 1080p. Motion levels: Medium, High. The still is the brief. The model moves it.

Five seconds is a short hold. Ten seconds is the long end of this row — not a plant documentary. Identity drift likes the long end. Until the firetube holds, stay on 5s. A 10-second miss is a long miss of a melted weir.

Medium motion on an approved shell is often the whole job. High motion on an unapproved package is ten seconds of drift and a new tube. Prompt the camera, not the process. "Slow dolly across the firetube, hold, no new objects." Stop. "Show rich glycol boiling down to 99.5%" is a different generate. It will not match the SKU. It will not match last week's overlay.

Because audio is false, this clip is silent. Do not write a VO into the I2V prompt and wait for a voice. Do not file a bug on a row that lists audio false. Score later, or do not add voice.

Do not morph tube diameter, burner count, or the still-column height. If the still shows a single firetube, the clip may not finish as a twin. If the still shows a horizontal reboiler, the clip may not stand up. Those are claims. A buyer will notice them before they notice your logo.

Water content is overlay, not Wan glyphs

Water content is a claim: lean glycol purity, sales-gas dew point, lb/MMscf, the number on the ticket you will stand behind. A 99.5 that becomes a 98 for six frames is a mis-ship. Video models redraw letterforms every frame. Text you cannot afford to get wrong does not live inside the motion.

Add a text overlay to a video burns one authored line for the whole clip — top, center, or bottom. That is the water number, a CTA, the package model. It does not transcribe. A mute reconcentrator has nothing to listen to. Overlay after picture lock. Swap the number for the next sample without regenerating the firetube.

If the names need to change at 0s, 3s, and 6s — lean purity, then dew point — that is timed overlays, not auto-captions of a silent shell. Crop before you overlay. Do not ask Wan to typeset "%" on the sight glass. Painted type is a miss you cannot edit when the lab number changes.

The T2V sibling is a bed, not this unit

Wan Video 2.5 Text to Video is the other Wan 2.5 row: no still required, 1080p, 15 credits, 5s or 10s, native audio. Same rate. Different object. It invents the frame. A silent I2V pass of your firetube is the identity job. T2V is a bed if you actually want a soundtrack and you are willing to say the picture is not the serial.

Write the mix on T2V if you use it — burner hiss, a quiet pad — because a picture-only prompt still returns a soundtrack the model chose. Do not paste a T2V prompt into the I2V row and hope. Do not use T2V to "match this reconcentrator" from a sentence. A sentence is not a plate.

Use T2V only when the unit does not have to be yours. The moment the firetube must match a catalog photo, T2V is the wrong door. A constructed operator walking "what water content you should spec" is a talking head you do not need. Keep lips off the skid. Overlay the number.

FAQ

Can I text-to-video a typical TEG reconcentrator instead of uploading mine?

No. A sentence is not a plate. T2V invents a cousin firetube. Upload the unit you ship to Wan Video 2.5 Image to Video and describe only the camera.

Does I2V add native audio if I ask for a plant bed?

No. Audio is false on this slug. The clip is silent. If you actually want a bed, that is the T2V sibling at the same 15 credits, native audio, and it will not hold this serial. Score the I2V plate later, or write the mix on T2V knowing the picture is invented.

Where does water content go if it cannot be in the prompt?

Type it. Overlay after picture lock. Lab numbers change; the firetube should not. Prompted glyphs are a miss you regenerate to fix.

Why not 10s every time so it feels like a plant film?

Ten seconds is the long end of 5s / 10s, not a facility documentary. If the firetube is wrong at 5s, it is wrong longer. Iterate short. Overlay the water number on the keeper.