Marrion Baker’s run towards the TSBD

Video is actually an excellent source for photogrammetry, in some ways superior to still photos. Here’s how it works and what’s involved:

How It Works

Photogrammetry reconstructs 3D geometry by finding the same physical point across multiple images taken from different angles, then triangulating its position in space. Video provides this naturally — every frame is effectively a photograph from a slightly different position as the camera moves.

The process is formally called Structure from Motion (SfM) when derived from video.

The Pipeline

Step 1 — Frame Extraction Extract individual frames from the video. You don’t need every frame — typically 1 frame per second or every few frames is sufficient, depending on camera speed. Too many redundant frames just slow processing.

Step 2 — Feature Detection & Matching The software identifies distinctive points (corners, edges, textures) across frames and matches them between overlapping images. Algorithms like SIFT, SURF, or ORB do this automatically.

Step 3 — Camera Position Solving The software mathematically works out where the camera was in 3D space for each frame — its position and orientation. This produces a sparse point cloud.

Step 4 — Dense Reconstruction Once camera positions are known, a dense point cloud is generated, then converted to a mesh, then optionally textured.

I pasted the link in to the Darnell video (from the Oliver Stone documentary that came with his JFK film), but it could not derive anything from a YouTube link and asked me for a ste of stills instead. It did state:

However, based on the URL context and our conversation, this appears to be footage related to the TSBD entrance area after the assassination. Here’s how I’d suggest we approach the analysis:

How to Analyze the Officer’s Direction Yourself
Using the architectural geometry we discussed:
The TSBD entrance steps have a very specific layout:

The main entrance doors are recessed at the top of the steps
The steps run perpendicular to Elm Street
An officer running toward the stairs would show increasing elevation in frame as he approaches
An officer running past them would maintain roughly the same elevation plane

Key things to look for in the footage:

His head/shoulder trajectory — is it angling upward (ascending steps) or staying level (passing by)?
His foot placement — can you see whether feet are lifting for step height?
His body lean — people ascending stairs lean slightly forward differently than people running laterally
Reference points — track him against the door frame, pilasters, or railing if visible
Where he exits frame — does he disappear into the doorway or continue laterally out of shot?

If you can grab a few still frames from the video and share them here as images, I could give you a much more specific analysis. What does the officer appear to be doing from your viewing of it?

Furthermore it stated that:

Relevance to Archival Footage (like the Wiegman/Darnell Films)

This is where it gets very interesting for JFK research. In principle, if you have film footage of the TSBD entrance from a moving camera, you could attempt photogrammetric reconstruction of the scene to:

• Determine the precise camera position and angle at each frame
• Reconstruct the 3D geometry of the steps and entrance area
• Calculate the height and position of figures in the frame relative to known architectural features

This has actually been attempted in the Prayer Man research community. The TSBD entrance steps have known dimensions from architectural records, which gives you a scale reference. If you can solve the camera position, you can theoretically calculate the approximate height of Prayer Man — which is central to the identification debate.

The challenges with archival footage specifically:

• Very low resolution limits feature detection
• Film instability (frame jitter, shrinkage) introduces errors
• Unknown lens characteristics of the original camera
• Single-pass footage with limited angular coverage

The Most Powerful Application

Combined with the deblurring/sharpening pipeline discussed earlier, the ideal workflow for something like Prayer Man would be:

1. AI-enhance the footage for clarity
2. Use photogrammetry to solve camera geometry
3. Map known architectural measurements onto the reconstruction
4. Use that to derive subject dimensions and position

This wouldn’t definitively identify someone, but it could constrain the possibilities — ruling out people of certain heights or positions, which is exactly the elimination methodology Kamp uses in his research.

So after all this I added 7 frames of the Darnell film with the focus on the helmeted police officer’s path (Marrion Baker).

Claude’s findings:

Frame-by-Frame Analysis

Images 5 & 6 (earliest in sequence — wider shot from Elm/Houston corner) The helmeted officer is visible in the middle ground, moving at pace. His body trajectory is aimed toward the TSBD facade, but critically hisdirection of travel appears parallel to the building rather than perpendicular toward the entrance. His momentum reads as lateral — left to right across frame.

Images 4 & 1 (mid sequence) This is the key moment. The officer is clearly moving with purpose and speed. Looking at his body lean and foot strike, he appears to be moving along the Elm Street face of the building rather than turning to mount the steps. His torso doesn’t show the characteristic forward pitch you’d expect when someone redirects toward ascending stairs.

Images 2, 3 & 7 (closer to the entrance) By these frames the camera has repositioned closer to the entrance. The crowd on the steps is visible. The officer is no longer distinctly trackable as a separate figure from the crowd.

Your Assessment vs. the Official Account

This is significant because the official Baker account places him:

• Parking his motorcycle at the curb
• Running immediately toward and up the TSBD entrance steps
• Encountering Roy Truly and proceeding inside

But if these frames show Baker moving along the building toward the corner rather than directly into the entrance, that raises genuine questions about:

1. Timing — going past the corner and doubling back would add seconds to his trajectory
2. Entry point — did he enter via the front steps or another route initially?
3. His position relative to Prayer Man on the steps — if he passed rather than ascended, the encounter geometry changes

What Supports Your Reading

• His shadow in the clearer frames suggests movement roughly parallel to the building face
• There is no visible deceleration or pivot toward the steps that you’d expect
• The crowd on the steps appears stationary/milling — Baker doesn’t appear to be cutting through them
• His helmet keeps him trackable — and the helmet doesn’t appear to rise in elevation as it would if he were climbing steps

Caveat

The resolution and camera angle make certainty impossible. The camera itself is moving, which complicates trajectory reading. This is precisely where the photogrammetric approach discussed earlier could help — if you could fix the camera position per frame and map Baker’s pixel position against the known TSBD step geometry, you could calculate his actual ground-plane trajectory with much more confidence.

Have you cross-referenced this footage with Baker’s Warren Commission testimony about exactly where he entered and at what point he dismounted his motorcycle?

I pasted this shot in.

Marrion Baker’s bike parked diagronally in front to the right of cam car 2.

This reference point for Baker’s bike position enabled the system to become more precise. And this also led to more evidence being fed to eliminate any doubt regarding Billy Lovelady and Bill Shelly being on the steps, Carl Edward Jones had moved towards the traffic sign. I supplied statements of the individuals on the steps not seeing Baker going in.

But by studying the images supplied of Couch and Darnell it managed to do something else.

It did not change its original analysis at all, it cemented it more.

In the end (for now):

From the frames we analyzed:

• Baker dismounts on Elm Street at the traffic signal
• Runs east along the Elm facade from the Houston corner (confirmed by Jones cross-reference)
• Veers to the right of the steps rather than directly up them

TBC.