Cannot open some .NEF files. “invalid vector subscript”

  Some of the .NEF files in this directory will open, and some will not.  The attached .ZIP file contains one that will not.

  This is a picture taken, an hour or two ago, with a Nikon D3200.

  I’ve never before had any difficulty opening .NEF files from this camera, in Photo AI.  I just barely got home, and am now trying to process these images.

DxDiag.txt (144.5 KB)
Screenshot 2024-10-13 222717


Topaz Photo AI [v1.x.x] on [Windows/Mac]
DSC_1691.zip (16.8 MB)

  For now, worked around the problem by opening the .NEF in Zoner Photo Studio (ZPS), and then saving it as a .PNG, and proceeding from there.  An extra step which ought not have been necessary, has never been necessary before.

  Here’s my final result…

  This is the comet that everyone is talking about, Comet C/2023 A3 Tsuchinshan-Atlas.  I took this picture at about 19:45 PDT, on 13 October 2024, at a location oddly-named El Camino de Jerry.

I gave the raw file a try on my system and got the same error in TPAI 3.2.2, but didn’t get an error in the old beta version 3.2.0.3b.

I also found that the new Gigapixel 8 beta opened the file fine.

Awesome image @Bob_Blaylock!

I was able to open this file on my machine and process the NEF to a DNG file.

I want to see if updating your GPU driver will resolve the problem, as there is a more up-to-date driver that has been released.

Please use the following link to update your driver:

After the update, let me know if you are still experiencing the problem.

  I updated the driver, and I still get exactly the same result.

I’d like for our team to take a look at this issue together would you please send us an email at help@topazlabs.com and the TPAI team will work together to assist :slight_smile:

@john.topazlabs

  On taking a further look, I seem to have figured out a common parameter distinguishing those files that the current TPAI cannot open, from those that it can.

  My camera is a Nikon D3200, c. 2012-2013. I have the one modern “kit lens” that came with it, and a few very old lenses, from the late 1960s to early 1970s.  I usually use the modern lens, but on occasion, I use the very old lenses.

  For the very old lenses, I have to put the camera in manual mode.  Of course, these old lenses, not having any electronics in them, do not report anything back to the camera regarding their own parameters or settings.  The EXIF data for pictures taken with these lenses, unless I edit these data otherwise, contain bogus values for parameters related to these parameters and settings; for example, reporting the aperture as ƒ/0.0 or the focal length as 0.0 millimeters.

  It appears to be the case that the current version of TPAI produces this error when I try to open a .NEF file that was taken using one of my old lenses.

  The only difference between a .NEF file taken with one of these old lenses, from one taken with the modern lens, as far as I know, should be that the EXIF data pertaining to the lens itself would not contain valid information.

The attached .ZIP file contain three .NEFs.

  • DSC_1711.NEF was taken with the modern lens, in automatic mode.
  • DSC_1712.NEF in manual mode was taken with the modern lens.
  • DSC_1713.NEF was taken in manual mode using my 1972-vintage NIKKOR-S Auto ƒ/1.4 50mm lens.  It is this lens that I used, a few nights ago, to take pictures of the comet; and the .NEF files generated thereby, that I found that TPAI could not open

  I don’t use the old lenses very often, so I don’t know how long this issue has been present.

  I did look at a directory from earlier this year, containing pictures taken with a different old lens, and found that the current TPAI cannot open those, either.

  File file Helicopter.zip contains a .NEF file DSC_9569.NEF taken in March of this year, using my Vivitar ƒ/3.8 85-205 millimeter zoom lens, made in the late 1960s or early 1970s.  Whatever version of TPAI I was using then, was able to process it just fine, but the current version cannot, and gives the same error I get when I try to process my comet pictures from a few evenings ago.

TopazPhotoAIError_20241015.zip (56.2 MB)

Helicopter.zip (20.7 MB)

@john.topazlabs

  One other thing I have realized, that might be relevant.

  as I already noted, the only difference that I believe there should be between a .NEF file taken with a modern lens, that the current TPAI can process, and one taken with one of my old lenses, that the current TPAI cannot; should be that the former would contain valid EXIF data reported by the lens, while the latter would not, since the old lenses don’t report any data, and would instead contain bogus values in any lens-related data that are recorded in the image file.

  My camera is usually set to create, for each picture that I take, a .NEF file, and also a .JPG file.

  The .JPG file that corresponds to a .NEF file that the current TPAI cannot process, ought to contain the same EXIF data that are in the .NEF file.  TPAI is still able to open and process that .JPG file, even when it cannot open the corresponding .NEF file.

———

  And one other thing that could be of some relevance is that in July of this year (2024), I bought a new computer. Before, I was running a seriously obsolete computer, that actually falls fairly far short of what you claim the minimum system requirements to be for TPAI.  It was able, in my experience, to run TPAI just fine, albeit very slow.  That is the computer that I was using when I last successfully used TPAI to process .NEF files taken using my old lenses.  The old computer is still up and running, set up in a way that with some inconvenience, I can still use it.  I suppose, if it is worth trying, I could update TPAI on the old computer, and see if the current version of TPAI running on it can process these files that the new computer running the current version of TPAI cannot.

@john.topazlabs

  I just now did as I said in the last post; updated TPAI on my old computer, and tried to run it on the same .NEF file that I included in Helicopter.zip.

  Same result.  It won’t open it, and gives the same error as on my new computer.  So, it’s not the computer; it’s something about this version of TPAI, compared to whatever version of TPAI was current as of the time I last was able to process that file successfully.

———

  I do note that a different issue that I reported back on 11 July 2024, and which at least two other users have also reported, still occurs on my old computer.  That one appears to be computer-specific; it went away when I switched to my new computer.  A bizarre user interface issue where everything that TPAI puts on the screen is shown 22 pixels below where it really is, so every mouse click registers 22 pixels above where you mean to click if you go by what is displayed on the screen.

I got the same vector subscript error on my pc with 3.2.2. I also checked and have no problems loading arw’s from an a6700 shot with no lens (rolling shutter tests), so f=0 and 0mm lens isn’t inherently the problem.

@john.topazlabs

  Seeing that there were a total of four different choices under that setting, I tried each one.  Same result, each time, whether I have Auto, CPU, NVIDIA GeForceRTX 4060, or AMD Radeon(TM) Graphics selected.

  The attached .ZIP contains an updated DxDiag.txt file, and the log generated by thsi run of TPAI.

Screenshot 2024-10-15 122447
20241015_122312.zip (49.4 KB)

  It appears, based on one test, that this bug has been resolved as of the new 3.3.0 version.

1 Like

  On the other hand, 3.3.0 introduced some much worse problems, so I’m back to using 3.2.2 until there’s a 3.3.? that can again process batches of images.

  So, @john.topazlabs, NOT RESOLVED until you have a USABLE newer version that doesn’t have this bug.  It is no fix to solve one bug, if you introduce an even worse bug in its place.

Could you provide any more information about the new bug you are experiencing? I will be sure to share this information with the developers.

  I did, in the 3.3.0 released thread.  I’ve gone back to using 3.2.2 for now, until it is fixed.  No more information, because I’m not going back to 3.3.x until this is fixed.

  If I load up a batch of images, and try to process them all, the processes are only done to the first image in the batch.  It pretends to process all the others, and saves new versions with names that indicate the processes that it pretended to do with them, but the new images are unchanged from the original images.