In this project, I am going over the process of scanning my entire photo and video archive, which sits at around 22 Terabytes, using ACDSee Photo Studio Ultimate 2026.
Acdsee Photo Studio Ultimate:
As an affiliate of ACDSee, I earn from qualifying purchases!
To make this computer accessible and feel more like a server, I am using virtual desktop software.
The server (on the ACDSee PC):
The client (on your main PC):
An issue with VNC is that it needs an active screen. You can leave your monitor running when accessing the computer, but another option is a virtual display driver:
ACDSee has sent me codes for this program in the past, but they have no relation to this specific project. I was simply curious to see how the AI keywords and face detection features would function when processing and using a large archive. As it turns out, it was quite the process getting this working and fulling scanning my ~22TB archive.
 |
| My local setup using a small dedicated monitor to check the server computer locally. |
My goal here was to create a client-server environment. Rather than running this compute heavy process on my main computer, I wanted to offload the work to a separate machine with one of my archive copies attached. Obviously, scanning this volume of data takes a long time, so separating it from my main computer seemed like the best approach.
The Software Setup: TightVNC and Remote Ripple
To make this setup work, I needed some specific software to manage the archive serving computer remotely.
The first piece of software I used is TightVNC. This acts as the server application on the archive computer. It allows me to connect from other computers to that one, giving me full access to the desktop. It is decently fast and functions well enough to access ACDSee itself or move files around.
 |
| The Remote Ripple interface allowing me to connect to the photo archive machine. |
On the client side (my main computer), I used Remote Ripple. This is the viewer application that lets me see and interact with the remote desktop running TightVNC.
Support Information: Virtual Network Computing (VNC)
VNC is a graphical desktop-sharing system that uses the Remote Frame Buffer protocol (RFB) to remotely control another computer. It transmits the keyboard and mouse events from one computer to another, relaying the graphical screen updates back in the other direction, over a network. TightVNC is a popular free and open-source implementation of this system, designed to work well even on slower network connections.*
Network Configuration
A crucial detail for this setup is networking. You will want to assign a fixed IP address (Static IP) to the computer hosting ACDSee (your server). If you leave it dynamic, the router might assign it a different address after a reboot, breaking your saved connection settings in Remote Ripple.
You will have to learn a bit of networking to assign a static IP via your router settings, but it ensures that the number associated with that computer never changes.
 |
| The server computer running ACDSee locally on a small external screen. |
The Monitor Problem
I ran into a specific quirk regarding displays. I have a tiny screen attached to the server computer for local maintenance, but I do not want that screen powered on and running 24/7. The issue is that TightVNC needs a display that it can push to connecting users.
To solve this, I installed a Virtual Display Driver. I found an open-source one on GitHub (I used the "Virtual Display Driver"). This creates a fake screen in Windows, allowing the remote desktop software to "see" a desktop even when the physical monitor is turned off.
Stability Hurdles
Once the computer was set up, I started scanning the archive. Immediately, I ran into issues. The program would run for a few minutes or hours and then crash.
After some troubleshooting, it turned out to be a hardware stability issue, likely related to the computer's RAM. It seems like ACDSee needs a rock-solid system to get through the cataloging process. I recommend scanning your RAM with tools like MemTest86 to verify it is good before attempting this. Even after stabilizing the hardware, there is always a chance ACDSee might crash during such an intensive process, so you have to monitor it. I think it happened once or twice for me.
Configuring ACDSee for a Large Archive
Once connected via Remote Ripple, I configured ACDSee to focus entirely on scanning the files.
 |
| The Options menu in ACDSee where Face Detection settings can be toggled. |
In Tools > Options, I made the following adjustments:
- Face Detection: Enabled. I set it to automatically scan while browsing.
- AI Keywords: Enabled. I allowed the indexer to access and use this feature.
- ACDSee Indexer: I set this to run when the computer is idle, with a delay of just one minute so it starts working quickly.
 |
| Configuring the Indexer to specifically target the photography and videography folders on the external drive. |
I specifically assigned the two main folders on my external drive to the Indexer so it wouldn't waste resources scanning the Windows drive.
The Cataloging Step First
Before letting the AI indexer run, you should use the Catalog Files feature. If you don't, the program won't know of any files to process unless you manually browse every single folder.
Go to Tools > Database > Catalog Files.
Important: I do not recommend enabling thumbnail creation for an archive this size. It is incredibly slow. It just didn't make sense, and the thumbnail process does not perform face detection or AI keywording that I want. I disabled thumbnails and set the file formats to include images and videos.
 |
| The Catalog Files dialog. Note that "Build and include thumbnails" is unchecked to save time. |
The Data: What's in 22TB?
With such a large archive, accessing the Dashboard features that compile statistics can be slow. However, seeing the data is interesting! Here are a few screenshots of the The ACDSee Dashboard overview showing information on the database.
According to the database statistics:
- Most Used Camera: Canon EOS M50. (I owned two of them at one point).
- Total Cameras: 111. (A nice number).
- Total Files: 345,882 images and videos. This actually feels low for how long I have been shooting, but it is what it is I guess.
- Favorite Format: RAW.
- Resolution: 24 Megapixel. This seems to be the sweet spot for me; higher resolution cameras are significantly more expensive a lot of the time besides the Canon EOS M6 Mark II bodies I used at one point.
 |
| A Nikon 70-210mm F4 lens, one of the many lenses cataloged in the database. |
Testing AI Keywords
The main reason I did this was to test the AI Keywords and Face Detection.
One limitation of ACDSee is that there is no easy way to see a master list of all found AI keywords and their quantities. You have to guess what might be there or look at individu
al files to see what tags were applied. I wish they would add a feature to list them all. I searched for simple terms like "computer" in the AI Keywords field.
 |
| The search results. |
It found a massive amount of results, including photos from the Vintage Computer Festival Midwest. It was generally accurate. I also searched for "glasses."
 |
| Search results for "glasses." It successfully detected glasses on faces and standalone pairs. |
It found a huge number of photos of me wearing glasses, but also found glasses sitting on tables by themselves. This is could be useful if you are looking for photo examples of specific objects that the AI keyword system knows about.
I also searched for "car." It found plenty of cars, highway shots, and even some things that were questionable like a horse drawn carriage, but close enough.
Face Detection and Hardware Limits
The People tab is where the hardware limitations of my old Intel i7-6700 CPU based computer really showed up.
 |
| The People tab showing detected faces. |
The software picks up a lot of things. It finds people, obviously, but also illustrations of people and occasional false positives like lens caps. You can adjust the detection sensitivity settings and rescan to mitigate this.
The fastest way to organize is to look at "Unnamed" and "Ungrouped" faces. However, there is a Group feature.
 |
| Task Manager on the server PC showing the CPU pinned at 100% utilization during the grouping process. |
The grouping process is very CPU intensive. On the older computer I used for the server, the CPU utilization hit 100%, and it took multiple minutes just to load the grouping interface. ACDSee does not seem to cache this grouping data, so it re-scans ever time I try to use it.
When it finally worked, the grouping was decent. It seems to group heavily based on the photoshoot or specific context. For example, it found a specific person at a sporting event and grouped those photos together. It also successfully identified me across different contexts and thumbnail styles, regardless of background color.
 |
| The results of the grouping feature, showing the same person identified across different images. |
Conclusion
This was a fun project, and the client-server approach is viable, but there are bottlenecks.
The CPU I'm using here is a major limitation. If you have a massive archive, you really need a high-end CPU to make the People Detection features usable day-to-day. Searching AI keywords is decent on older hardware but also somewhat slow.
For file transfers, I simply used Windows File Sharing. While I could transfer files through the VNC software, it is very slow. Windows File Share gave me transfer speeds between 50 and 100 MB/s over my network. It isn't as fast as plugging the QNAP TR-004 DAS drive array directly into my main computer, but it is functional enough for viewing and retrieving specific files.
If you plan to do this, make sure you back up your database after the long scanning process is complete!