Neural Networks in Practice: A Case Study from a Video Art Project for Darkly Art Magazine. Kirill Akishkin — Director & Photographer | Cinematic Visual Storytelling

From Photography to Hybrid Form.

The Witch Passion project was shot about a year ago as an authorial photoshoot. Initially, it existed as a photo-narrative series accompanied by video footage — a complete visual form with its own mood, rhythm, and world.

However, over time, it became clear that this material had not exhausted itself. It demanded continuation and reinterpretation.

It was at this point that neural networks entered the workflow — not as a tool for acceleration, but as a means of expanding the artistic language.

Neural Networks as an Artistic Medium, Not Automation

For me, working with AI has never been about “doing it faster” or “simplifying the process.” On the contrary, neural networks are integrated into my working method as a medium — on par with the camera, editing, or color grading.

I perceive AI as a space for dialogue with the image:

  • where the result is not entirely controllable,
  • where an element of unpredictability arises,
  • where the artist works not with a format, but with possible worlds.

In the Witch Passion project, neural networks allowed for a much more precise conveyance of the intended meanings through visuals, deepening the existing video sequence and amplifying the project’s emotional and symbolic dimension.

Between Photography, Video, and Video Art

During the work, the project lost its rigid genre classification. It is no longer photography in the classical sense, nor video in the conventional understanding.

It exists at the intersection of:

  • photography,
  • video,
  • video art,
  • and algorithmically generated states of the image.

Here, neural networks do not “draw for the author”; they transform the original visual material, preserving its DNA while shifting the emphasis — towards motion, distortion, and metaphysical tension.

The Practice of Working with AI in the Project

  1. Original Authorial Material — photographs shot specifically for the project’s concept.
  2. Interpretation of Images through Neural Networks — not copying or stylization, but a search for new states of the image.
  3. Manual Selection and Editing — the key stage where artistic decisions are made by a human, not an algorithm.
  4. Final Video Structure — rhythm, pauses, breath, visual dramaturgy.

It is important to emphasize: neural networks do not replace the author’s vision; they amplify it — provided there is conscious and critical engagement with the result.

Why This Is Important for a Contemporary Artist

Today, an artist works under conditions of blurred boundaries between media. Formats no longer define the statement — they are merely temporary shells.

Neural networks are becoming part of this process not as a trend, but as a logical continuation of artistic inquiry. They allow artists to:

  • reassemble archives,
  • return to old projects,
  • discover new meanings within them,
  • create hybrid forms previously impossible.

Witch Passion and Darkly Art Magazine

Darkly Art Magazine will feature the photo-narrative part of the Witch Passion project — photo art and a textual sketch — in the issue themed The Art of Mystery. For me, this is an important context; a magazine working with dark aesthetics, borderline states, and visual mysticism has proven to be an organic space for the project’s existence.  

Instead of a Conclusion

Working with neural networks demands no less responsibility than working with a camera. It is not a shortcut or a magic button, but a tool that lays bare the level of the author’s artistic thinking.

Witch Passion became for me an example of how past material can acquire a new form and new breath — if one allows technology not to replace the author, but to enter into a dialogue with them.

If this approach to AI, video art, and hybrid forms resonates with you — this journey is only beginning.

RU
EN