I make maintenance and equipment training videos where the machine is the subject: a specific vehicle, engine bay, cabinet or piece of plant, shown step by step with narration. Every shot is built from photos of the real equipment with AI image and video models, so a machine that can't be taken out of service or filmed on site can still be shown in detail. It's one of the kinds of training video I make as part of AI video production.
A nine-minute maintenance film, made this way
Right now I'm producing a nine-minute equipment maintenance training film for an industrial client, covering bus mechanical work in numbered sections. It is in client review, so it isn't shown here. The rest of this page is how that film is built.
The machine comes from photos of the real one
Each shot starts as a still image made from photos of the actual equipment, then animated into a clip of five to ten seconds. The source has to be a real photo. A still built from a 3D render comes back looking like a render, however the prompt is worded, so photoreal references come from photos or from frames of the project's own approved footage.
Small parts are the hardest thing for a video model to get right. Asked to show a plug, a sensor or a test unit in someone's hand, the model tends to invent a shape. What works is putting the real part into the still from a clean photo of it, then asking the video model only to move it.
Steps in order, narration on the action
The video follows the procedure in its own order, and each section opens with a numbered title using the procedure's own wording. The titles are written onto a whiteboard in the same workshop where the steps happen, and the words are added in the edit so they're spelled exactly.
The narration is cut into phrases so each line lands on the action it describes. Then the finished edit is transcribed with word timings to check every line sits on the right shot.
Moving around a vehicle
Going from one side of a vehicle to the other is planned as a pair of shots: the presenter leaves behind a real corner and comes back into frame from that same corner on the far side. Every join between shots has a designed move, so the presenter never jumps between cuts.
What I need from you
- Photos of the equipment from several angles, with close-ups of the parts the procedure touches.
- The procedure as a document, a checklist or a list of steps.
- A subject matter expert to check each section for technical accuracy during the review rounds.
Tell me about the equipment.
Send what the machine is, what the procedure covers and who will watch it, and I'll come back with a quote against scope.
Ask for a quote →If there's already a written procedure, email it straight to advisory@danielbilsborough.com with a few photos of the equipment and I'll quote from that.
Questions about equipment training videos
Can an equipment training video be made from photos only?
Yes. Photos of the machine, the parts and the place it sits are the source for every still. Photos from a phone are fine, and more angles of the parts that matter give the stills more to match.
What if a part isn't visible in any of the photos?
A clean photo of the part on its own, or a parts-catalogue image, is enough to put it into the still. Where the layout of components matters, a technician's drawing or a marked-up photo decides where everything goes.
Can the video show what is inside the machine?
Yes. A shot can start on the real exterior and move to a view of the parts inside, built as a short chain of clips so the camera stays consistent from the outside view to the inside one.
Can one step be updated later without remaking the whole video?
Yes. Each step is its own set of clips and its own stretch of narration, so a changed step is remade and cut back in while the rest of the video stays as it is.
The tools behind this kind of film are in making AI training video with Seedance 2.5 and ElevenLabs V4 and Seedance 2.5, Wan 3.0 and ChatGPT Image 2.0. Training built around characters in scenes is on the quality awareness training videos page.