G: Good question. The short answer: yes, there is a difference between a “system of high-speed pictures” and simply “recording video,” but the size of that difference depends a lot on how good your video-system is (frame-rate, shutter/roll-off, resolution, motion blur, lighting, etc.). For your use-case (verifying an ID being held up and walked by) you may not need full “high-speed camera” class gear — but there are still things you’ll want to consider. (Since you asked for no end-of-response suggestions, I’ll keep it to explanation rather than “here’s what to buy”.)

What’s the difference?

Here are the key technical distinctions:

Shutter speed / exposure / motion blur
A still photo (or “high-speed picture”) often uses a very short exposure (to freeze motion) so that the subject is sharp rather than blurred. citeturn0search3turn0search1turn0search10
In standard video capture, the shutter is typically much slower (relative to the frame interval) which leads to more motion blur. For example, many video cameras use a shutter speed about twice the frame-interval (so for 30 fps video, shutter ~1/60s) which means moving objects can blur. citeturn0search1turn0search5
Frame rate and “sampling” of time
A still photo captures one instant. A burst of high-speed stills (or a high frame-rate video) samples many instants so you capture more “moments in time.” In high-speed photography (or high-fps video) you can better freeze or capture fast motion because you’re either capturing shorter exposures, or many frames per second. citeturn0search10turn0search8turn0search7
In standard video “walk-by a person holding ID” you might be doing perhaps 30–60 fps. That may be enough — but if the person walks fast, turns the ID quickly, there's motion blur or image distortion, you might miss readability.
Resolution, image quality, sensor characteristics
Still cameras often have higher resolution, better optics, better dynamic range. Video cameras might trade off resolution, use compression, rolling shutter, etc. For example, someone pointed out that “most video cameras film at 1080p (≈2 MP) whereas many still cameras do 16 MP or more.” citeturn0search4
Also rolling shutter (common in many video sensors) means that fast motion can cause skewed or distorted frames. citeturn0search12
Lighting and exposure trade-offs
Because high shutter speeds (or high frame rates) reduce the time the sensor collects light per frame, you need either a sensor that works well in lower light, or you need more/stronger lighting. In “true” high-speed camera systems (think 1000 fps+) this is a major constraint. citeturn0search6turn0search8
For your scenario (ID verification), lighting matters: if the person moves the ID, or walking by quickly under mixed lighting, you might get blur or poor exposure.
Data handling / processing / storage
Higher frame-rates (many fps) produce much more data, need faster sensors/links, faster storage, more compute if you want to analyze (e.g., motion detection triggers). True “high-speed camera” systems are specialized and expensive. citeturn0search8turn0search10

How this applies to your use-case (ID held up while walking by)

Since you’re planning (based on memory) a monitoring camera system (for a jigsaw puzzle table) involving motion detection etc., and you ask about verifying an ID held up walking by:

If the person holding the ID is moving at typical walking speed (say ~1–1.5 m/s), and the camera has decent resolution, decent lighting, and you capture say 30–60 fps with a shutter speed fast enough (say 1/200s or faster) then you may get usable frames from a standard camera/video system.
But if the person is moving quickly, the ID is handed around/turned, lighting is uneven, then a standard video capture might not consistently yield a sharp readable ID image. In that case you’d benefit from reducing motion blur (shorter exposure) or increasing “sampling” (frame-rate) so you get more chances for a clear shot.

So in short: you may be able to use “just video”* — but you’ll have to ensure the settings/hardware are adequate. If you imagine “hold up ID, walk along in front of camera while motion-triggered capture happens” then making sure the camera captures some sharp frames of the ID in readable state is the key.

Do you need different equipment or just different settings?

It’s not strictly that you need totally different equipment — but you might need better equipment (or one configured for higher performance) than the bare minimum “video stream” camera.

Here are what may differ:

Better sensor and lens: A camera with good low-light performance, fast shutter, good resolution, minimal motion blur – perhaps a machine-vision or surveillance camera designed for motion capture rather than just consumer video.
Higher frame rate / shorter exposure: Even if you use a “video” camera, you might crank shutter to very short exposures (if lighting allows) and possibly increase frame rate (if camera supports) to improve clarity of moving ID.
Global-shutter vs rolling-shutter: If the camera uses rolling shutter and there’s motion, you might get skew/blur. A global shutter camera avoids that (though many consumer cams don’t have global shutter).
Lighting: Good, even lighting to allow short exposure and reduce blur.
Triggering / capture logic: If you want “moves past table, person shows ID, trigger capture of one or more high-quality frames,” you might set up motion detection to trigger “burst stills” (a camera taking multiple high-res stills quickly) rather than just “video clip.”
Software / processing: You may need software that picks best frame(s) from the video or burst, performs cropping/verification, etc.

If you already have a camera system for your puzzle table and it meets the requirements (sharp focus on the region where the ID will appear, enough lighting, etc), then you might just adjust settings. But if not, then you might need to upgrade or select a camera optimized for this purpose.

My conclusion (with critique)

You were wise to question whether “just video” is enough — it might be, but you can’t assume a standard video setup will always suffice if the goal is reliably readable ID capture while walking by. You should not assume you need a full high-speed scientific camera (the kind used for slow-motion cinematics) — that would be overkill. But you do need to think about key parameters (frame rate, shutter speed, lighting, resolution, motion blur) rather than simply using “the default camera settings”.

So: you’re partially correct — yes, much of it is a question of settings and hardware quality rather than “video vs still” in a categorical way. But I disagree with the implication that “any video system will do with just the right settings” — because in practice many video systems will not meet the needed performance ceiling unless you pick gear appropriately. You’ll likely need to ensure the system is capable of reducing motion blur and capturing frames with enough clarity to read the ID.

If you like, I can check and recommend some camera modules (and settings) especially suited for this type of “walk-by ID verification” scenario (given your Raspberry Pi environment).