DPL9 Important Feature request

DxO PhotoLab Windows topics
1

DPL9 automatic Subjects recognition works nicely : human , animals car etc. But …

In order to improve the power of DPL9 and because new cameras deliver bursts of 100 or 10000 photos per sec, min hours and day and the every days, it will be highly helpful for users to enable a suitable batch processing on a quantity of image.
On the on-going DPL9.01 it seems that only the mask of the 1ST image is calculated and then applied to all following images in the batch. This lead to a wrong result because each image of bursts is different (movments of the subjects or/and of the photographer) .

Thanks in advance for latest cameras users.

1 Like
2

Have you tried to save the mask-settings e.t.c. from that first “template” as a template and then use that template on a selection of similar pictures?

3

Thanks I will try . Good suggestion. CT

4

I have tested this a bit and it is not as smart and versatile as in Capture One “Match Look”. I hope DXO will pick up some ideas from this because this is the very foundation to build effective AI-workflows on. In the examples below you can see that the white balance and exposure and a few other parameters as well can be propagated by unique individual AI-analyses and actions for each and every target picture.

From I have been able to understand it works best in Photolab if we stick with the premade masks in AI-masking. Using the older masking tools and just save the settings to a preset did not look good when I tested it yesterday did not come out all that well.

Maybe some other users can falsify my testing and help me and others to more in detail understand what works and what is not.

5

What a job Sten ! Thanks I made the tests only with Subject AI ( animals) and i found the same: Using a setting enables to get the wright mask on each image of the batch. But sometimes the other settings are or are not managed under the mask: i would say after this primary test that it works for the Mask top level : all contours on each image have be been properly calculated . But at sub mask level the other settings that we may want to apply to the mask sometimes are applied sometimes not.

As all functions ( masking under AI) and other settings work separatelty and partially when launching a batch it should be easy for DPL9 to get the AI masking C1 level with just a féw days effort.

6

As I also have written a lot of users have severe problems using the predefined AI-masks since they cause crashes and Internal Errors that needs restarts to be reset.

People having these problems can’t use the predefined AI-masks at all and are far better off using the free hand hoovering method instead. That seems to be totally stable and is much faster and much more efficient too as long as just one picture is edited.

Though in order to use one or several predefined AI-masks in a template to save as your own AI-driven presets, we need to be able to use even the predefined AI-masks. For the moment that is not possible for many users including me and that has to be fix so they can be used when intelligent automatic AI-mascing is needed when we want to use presets like that to increase our productivity when batch updating many pictures in one selection.