Hello everyone,
I am not a specialist researcher of the Dyatlov case and I am new here, but I have been taking photographs for many years. Frame 34 interested me as a photographic problem: what real scene might produce these shapes, and what might the person behind the camera actually have been looking at?
This is a hypothesis about the photograph, not a theory explaining the group's deaths. I would like to show the reasoning and the stages of an AI-assisted visual exploration, rather than simply present a picture and say, "Look how similar it is."

The supplied copy of Frame 34 used throughout this exercise. Source: DYATLOVPASS.COM; this is a reproduced image, not an examination of the original negative.
1. The central polygon: a possible lens ghostThe relatively compact pale shape near the centre looks approximately octagonal to me. My first working interpretation was an internal lens reflection whose boundary is shaped by the aperture, rather than a separate physical object.
I used AI assistance to check the lens information and locate references. Eight-bladed versions of the Industar-22 do exist; some early versions have ten blades. Therefore an approximately eight-sided reflection is compatible with an eight-bladed example, although I have not established the blade count of the particular lens involved here.
An aperture-shaped ghost near the centre is optically possible with an off-centre source. However, we have not calculated or experimentally reproduced this exact reflection in an Industar-22. A separate defocused light could also acquire an aperture-shaped outline. The ghost interpretation is a working assumption, not an identification.
Lens reference:
https://radojuva.com/2013/03/obzor-industar-22-3-5-50mm-p/Ten-bladed early example:
https://radojuva.com/2020/07/industar-22-moskva/General explanation of aperture-shaped ghosts:
https://av.jpn.support.panasonic.com/support/global/cs/dsc/knowhow/knowhow15.html2. Why I questioned a long motion trailIf the whole extended formation on the left were the trail of a compact light moving relative to the lens during a long exposure, I would normally expect its associated ghost to move as well. The central shape remains comparatively compact, with a recognisable polygonal boundary. That made me wonder whether the left-hand formation largely existed at one moment, rather than being drawn entirely by movement during the exposure.
Initially, I assumed handheld night photography without a tripod, and suspected a short exposure with only slight camera shake, perhaps around 1/30 second or faster. I now need to distinguish that initial reasoning from what can actually be established: the photograph does not determine the shutter speed. Some ghosts move less than the main image, and a brief light event can be recorded during a longer exposure. Also, the case-file inventory mentions a camera with 34 exposed frames together with a tripod, so my initial assumption of no tripod is not secure.
The limited claim I retain is that the large left-hand shape need not be a long motion trail. Much of it could represent an extended scene blurred mainly by defocus.
Case-file inventory, sheet 5:
https://dyatlovpass.com/case-files-3-63. The uneven upper formationThe light above the main bright patch is uneven and lobed, with a darker indentation on the right. It reminded me of smoke illuminated from below. I treated the bright lower area and the upper formation as one related scene: a light source with illuminated material above it.
Smoke was my initial suggestion, not a recovered detail. Illuminated tent fabric, folds, steam or a combination of these could also contribute to the upper formation.
4. The roughly rectangular lower lightWhat particularly caught my attention was the lower bright area's approximately rectangular outline, with rounded corners. Strong defocus can soften a rectangular light source without making it perfectly circular.
When I raised this point, the AI suggested a bright rectangular opening, including a furnace or stove opening, as one possible explanation. We explored three variants: an incandescent opening with smoke, a rectangular lamp illuminating mist, and several closely spaced lights. The opening-and-smoke version is the one developed below.

AI-generated hypothetical scene: a bright rectangular opening with illuminated smoke. This version predates the later tilt and lens-rendering adjustments.
5. Introducing a small tilt and slight camera shakeWe then rotated that reconstruction about four degrees clockwise, lowering the right side. A separate processed version added defocus and a small diagonal displacement blend to represent slight camera shake. These are illustrative settings, not measurements recovered from Frame 34.

The same hypothetical scene rotated clockwise, before the additional defocus and shake.

Top: tilted reconstruction. Middle: octagonal defocus plus a small camera-shake blend. Bottom: the supplied copy of Frame 34. The central ghost was assumed and retained separately during defocus; it was not generated by ray tracing.
6. An approximate Industar-22 renderingWe also made an AI-generated version intended to reflect some reported characteristics of the Industar-22: softer bright boundaries, reduced contrast against an intense light source, and veiling glare. These characteristics vary with the lens version, aperture and individual condition. We then applied defocus and slight shake computationally, as before.
This is an approximate imitation of the lens's rendering, not a calibrated optical simulation. The working prompt assumed an eight-bladed version at f/5.6; neither that aperture nor the actual focus setting is known. Chromatic aberration was not independently calculated. The central ghost's position and brightness were assumed, not predicted.

Hypothetical reconstruction with an approximate vintage Industar-22 rendering.

Top: approximate Industar-22 rendering. Middle: the same reconstruction with added defocus and slight shake. Bottom: Frame 34.
7. Finding photographs of a stove replica afterwardsOnly after we had discussed the rectangular opening and generated the stove-like reconstruction did I find the following photographs on DYATLOVPASS.COM. The article also includes a stove-replica demonstration credited to Aleksandr Fedotov, February 2025.
Source article:
https://dyatlovpass.com/firewood-in-the-tent-ru
Contemporary tent-and-stove reconstruction photograph reproduced from DYATLOVPASS.COM, with source credit. This is not a photograph of the original 1959 scene.

A closer view supplied from the same article. The bright rectangular opening and illuminated tent fabric provide a concrete scene worth testing photographically. Source: DYATLOVPASS.COM.
The resemblance struck me: a rectangular fire-lit opening below, with irregular illumination spreading above it. In these modern photographs there are additional lights, so they are not a direct reproduction of the lighting proposed for Frame 34. Nevertheless, they suggest a practical test using the stove as the main light source, with the other lights switched off.
What this exercise does and does not showThe goal was to imagine a physically plausible scene, not to recover hidden detail or force every patch into an exact match. However, the generated images were explicitly guided by the original's positions and broad light contours. Their resemblance is therefore not independent evidence that a stove was photographed.
The source photograph's upper formation is more uniformly bright than our simulated smoke, and its lower bright patch is less regular. Its broad curved light boundary on the left also remains unexplained. Tent illumination, optical reflections and film or reproduction effects need to be distinguished rather than treated as interchangeable explanations.
There is also a historical constraint: Lebedev's testimony describes the stove in its case with the pipes inside it. My hypothesis does not establish that the stove was lit at the final campsite, or when this frame was exposed. It should not be used as evidence of a stove accident.
Lebedev's testimony:
https://dyatlovpass.com/case-files-313-315My question is narrower: could a bright stove opening and illuminated tent fabric and/or smoke, photographed out of focus, produce the broad structure of Frame 34?
I would welcome photographic objections, information about the actual stove and camera, and especially a practical test with an Industar-22. Reproducing the left-hand light formation and the central polygonal ghost together, rather than placing the ghost into a reconstruction, would be particularly informative.
All reconstructed scenes above are AI-generated illustrations. The subsequent defocus, rotation and shake processing were performed computationally. None of these images is recovered photographic evidence. For the illustrations, scratches and dust were omitted, and the thin angular line at the bottom right was treated as a possible surface defect; that classification has not been established from the negative.