IDOCInTheBox
v6.1
IDOCInTheBox
v6.1
Ophthalmologist of 39 years and self-taught AI imaging engineer. Author of Advanced Tape Restorer.
I have always taken things apart.
It started with the locks on the doors of our house. I disassembled them, worked out how the pins and the shear line did their job, and put them back together. Between about eight and ten I read the World Book encyclopedias straight through and Scientific American alongside them. I rebuilt bicycles for every kid in the neighborhood. Then I took apart lawnmower engines to learn how they worked, and went on from there — car engines, boat engines, mechanical and electrical appliances, and eventually computers, starting on a Radio Shack TRS-80 in 1976 and going on through systems, networking and servers.
None of that was a plan. It was simply the only way I have ever been able to understand something: open the housing and look.
I was a BS/MD student, which meant completing both degrees in a combined six-year program. The compression packed in a great deal of physics, chemistry and calculus in a short time, and it left me with a habit I still have — looking for the relationships between fields rather than staying inside one.
I have now practiced ophthalmology for 39 years. It is a specialty built on applied optics and image interpretation, so the curiosity found an obvious outlet. I learned the instruments as thoroughly as the medicine: the lasers, the diagnostic devices, the imaging systems. In time I could disassemble and repair most of the equipment in my own practice, and manufacturers' engineers began seeking me out at trade shows to talk through how their systems actually behaved in clinical use.
Some of that turned into real collaboration. I reported several pre-release software and interface defects on Carl Zeiss Meditec's IOLMaster 700, and I beta-tested and contributed design feedback and ideas — without charge — to their AI-based IOL calculation software. Zeiss sent a developer and two UI specialists from Germany to my office to watch how I actually used the program and take the ideas back with them.
Advanced Tape Restorer came out of exactly the same instinct.
I had started restoring photographs, then film, then video. What I found was a toolchain that was cumbersome and outdated — a different application for every stage of the job, and every handoff another chance to lose detail that is difficult to acquire and impossible to get back. There were too many small decisions to get right, and getting any one of them wrong quietly degraded the result. The failures had nothing to do with the source material and everything to do with the tools.
So I set out to build the unified tool I could not buy. As AI image enhancement matured I became interested in what it could do for this problem, and taught myself AI imaging and LLM engineering in order to bring it in properly rather than bolt it on. I had always used NVIDIA hardware for my other imaging work because it gave the best and fastest results, and I built ATR on the same foundation. When modern computers dropped FireWire and stranded the entire DV tape format, I designed and built a working capture device prototype and verified it against ATR itself.
I still look for new techniques most days. Increasingly I find myself asking how they transfer back the other way — into ophthalmic imaging, and into surgical guidance. The registration, noise-floor recovery and detail reconstruction that ATR performs on a damaged videotape are, mathematically, very close to the problems I have spent a career looking at through a slit lamp.
That is the next thing I intend to build.
Charles William Titone, MD
Founder, IDOCinTheBOX Creations LLC
Home video from the analog era is on a clock. VHS, Hi8, 8mm, Betamax and DV tapes lose magnetic signal every year they sit in a box, and the decks that can read them are going out of service faster than the tapes are failing. Most people have exactly one copy of footage that cannot be reshot.
Advanced Tape Restorer exists to get that footage off tape and make it watchable again. It combines a full hardware capture path — DirectShow analog capture, DV/FireWire stream copy, deck transport control — with a 20-engine AI restoration suite covering upscaling, frame interpolation, denoising, deblurring, face restoration, colorization and audio enhancement. Every restoration decision is measurable: the app reports VMAF, PSNR and SSIM against the source rather than asking you to take improvements on faith.
The design constraint that shapes the whole product is that the complexity stays hidden. VapourSynth pipelines, ONNX model conversion, VRAM scheduling and colorspace handling are real engineering problems underneath, but a person restoring their family's tapes should never have to learn any of it.
| Legal name | IDOCinTheBOX Creations LLC |
|---|---|
| Trade name (DBA) | IDOCinTheBOX Creations |
| Entity type | Limited Liability Company (LLC) |
| State of formation | North Carolina, United States |
| Registered with | North Carolina Secretary of State |
| Date formed | March 16, 2026 |
| Founder | Charles William Titone, MD |
| Principal product | Advanced Tape Restorer v6.1 |
| Website | idocinthebox.com |
| Contact | [email protected] |
Advanced Tape Restorer has been in continuous development since November 2025. Development predates the formation of the LLC — the company was formed in March 2026 around a product that was already well advanced.
Two builds have been published. The rest are internal development versions — feature work, testing and bug fixes versioned in a private repository and never released. They are listed because the engineering record is worth showing, not because they were downloadable. Full notes for every version are on the development log.
The current commercial release. Adds the full-frame stabilizer work — a widened temporal donor search that hands back roughly 4% of frame area, and border-artifact fixes across both full-frame policies.
The first commercial release. 20-engine AI suite, community plugin system with three pipeline hooks, scene editor, batch preset comparison, Story Intelligence, Visual QA, integrated music library and the full hardware capture module.
Planner and candidate-evolution work, expanded restoration engine coverage, and a long run of point builds through 5.0.10 refining stabilization and quality scoring.
Face reference library, colorization and audio enhancement pipeline expansion.
Quality metrics suite — VMAF, PSNR and SSIM measurement against source.
ONNX / NPU acceleration with DirectML, enabling 4K processing on 8 GB GPUs.
The free Community Edition, MIT licensed — QTGMC deinterlacing, BM3D denoising, VHS artifact removal, color correction, hardware capture and the full set of output codecs, without the AI suite. Still available on GitHub.
First working build of the restoration pipeline.
Support, licensing and business enquiries: [email protected]. For purchase or license-key issues, include your order ID and we will get to you faster.
Legal documents. End User License Agreement • Privacy Policy • Third-Party Licenses