Announcement

Collapse
No announcement yet.

ACDSee Pro Puzzles and Questions

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • ACDSee Pro Puzzles and Questions

    I have invested in ACDSee Pro 2022 and generally after a few struggles am beginning to like it. However as I have browsed these forums a number of questions have come up..

    1. If I regularly back up the database to a secure location, and I don't store any ACDSee data in the image file. What is the benefit of embedding ACDSee custom data in the image file which is what some posters suggest is a wise move ?

    2. I note that buried deep in these forums is the fact that if the "Face Recognition" function does not identify a face in an image, and you manually draw a box around the face and name it, ACDSee ignores the fact and does not use the image as part of it's face recognition data.

    3. Does ACDSee have a reporting function that will allow you to identify all the images that have had the faces manually added and identified. So that when a smarter better face recognition program is developed they can be added to ACDSee's face recognition repository.

    4. As many people have pointed out no software is bug free, how often does ACDSee release bug fixes and minor updates. Most of the software I use Topaz, DXO, Photoshop are distributing fixes and updates fairly regularly.

  • #2
    Originally posted by Dronenz View Post
    If I regularly back up the database to a secure location, and I don't store any ACDSee data in the image file. What is the benefit of embedding ACDSee custom data in the image file which is what some posters suggest is a wise move ?
    Asset management involves the use of keywords, captions, categories, notes, labels, ratings etc that enable you to search for, display, group on etc
    You can just have these in the database, and rely only on backups of the database should there be a problem with the current database. But databases can become progressively untidy and your most recent backup could already contain issues. There is also the issue of the possible loss of any records added after the last backup was made.
    .
    A much safer setup is to also embed the above metadata in the images themselves (or in the case of RAW images, in xmp sidecar files). You are then able to completely construct a new database by reading back the metadata from the images themselves. You choice entirely, but I'm one who strongly recommends embedding the metadata in the images.

    I note that buried deep in these forums is the fact that if the "Face Recognition" function does not identify a face in an image, and you manually draw a box around the face and name it, ACDSee ignores the fact and does not use the image as part of it's face recognition data
    I think you maybe misunderstanding the situation. As I see it, there are two parts to facial ID. The first is whether a face in an image meets with the criteria for it to be automatically recognized as a "face". The second is then matching that face to a name. If a face is automatically detected, then it can hopefully be matched against previously identified faces of the person.
    There will however always be photos where the face does not meet the criteria for automatic recognition (It may be for instance that the person is side on to the camera, is wearing dark glasses, or is just to blurred), but you know who the person is, and you want to the person to be identified in the photo. So you can manually add the face "bounding box", and you can name it. The face will show in people mode, and the name will show when you view the photo in View mode (when show face outline is turned on). But since this "face" didn't meet the criteria for auto recognition as a face, it isn't used as a reference to automatically identify other faces.

    Does ACDSee have a reporting function that will allow you to identify all the images that have had the faces manually added and identified. So that when a smarter better face recognition program is developed they can be added to ACDSee's face recognition repository.
    You can identify images where the faces were automatically identified. In Manage mode, select the Catalog Tab, scroll down to the Special Items (last section) and select Auto named faces.
    If you specifically want to identify just manually identified faces, you could first select the "Auto named faces", use Ctrl + A to tag them, then tag them. You can then go to the People section of the Catalog tab, use Click then Shift +click to group select all of the people, the set the filter to Un-taggeg..
    Remember to clear any existing tags, and embed any metadata in images that have the embed pending flag set BEFORE you do the above, and then clear all Tags and clear the embed pending flags once you have completed the above.

    As many people have pointed out no software is bug free, how often does ACDSee release bug fixes and minor updates. Most of the software I use Topaz, DXO, Photoshop are distributing fixes and updates fairly regularly.
    As far as I am aware, there is guarantee that there will be any maintenance releases. Some years there have been one or maybe two, and there have been updates to just the RAW library, but there was no maintenance releases at all for the 2021 version..

    I too use Topaz, and wish they would issue far less updates, but get them right.
    Last edited by Greyfox; 05-18-2022, 10:53 PM.

    Comment


    • #3
      Thaks Greyfox. The face recognition issue is really interesting, when I started playing with ACDSee a few days ago I loaded 30ish JPG images, there were a number of groups of 5 or 6 images that were identical but they had been colour modified, colour toned and B/W. Only about 50% of the images in each group were identified as "faces" despite the fact that they were ostensibly identical apart from subtle colour shading. Which is why I thought ACDSee had not loaded the images, they were in the database but not as "faces". Professional failed in the same way.

      Comment


      • #4
        Originally posted by Dronenz View Post
        Thaks Greyfox. The face recognition issue is really interesting, when I started playing with ACDSee a few days ago I loaded 30ish JPG images, there were a number of groups of 5 or 6 images that were identical but they had been colour modified, colour toned and B/W. Only about 50% of the images in each group were identified as "faces" despite the fact that they were ostensibly identical apart from subtle colour shading. Which is why I thought ACDSee had not loaded the images, they were in the database but not as "faces". Professional failed in the same way.
        I am mostly quite happy with the success rate of ACDSee Ultimate's Facial recognition, but every so often it does throw up something interesting.
        I have had a few situations where with 24 bit color scans of very old black and white photo prints, ACDSee has been able to both auto outline the faces, and also correctly identify them, despite there being a significant amount of noise in the images, and has been able to do the same with direct copies.

        But when those copies had Topaz DeNoise AI remove some of the noise (without visually removing any significant facial detail) ACDSee could not even auto outline the faces.

        When instead I used ACDSee Ultimate's Edit mode noise reduction (both despeckle, and luminance noise reduction) which visually did appear to soften the faces more than the Topaz app, ACDSee could still auto outline and identify the faces. These were full front on faces..

        Life is an endless series of puzzles

        Comment

        Working...
        X