Why not more *AI* technology?

[Guest]

Would be great if the Affinity suite had more AI based tools. For example:

1. Image upscaling like: https://letsenhance.io/
2. Remove BG of an image: https://www.remove.bg
3. Mix two photos into a new one: https://www.artbreeder.com/  (it's not just blending, but more like breeding the subjects of the photo):
4. Automatic scene labeling like this. To have rough masks to start working with. + Automatic depth map creation for inserting Depth of field

Edited March 25, 2020 by Guest
had a new idea

[Alej]

Woah, I especially like the scene objects highlighting (because I hadn’t seen it before like this).
Pretty sure the Serif guys are working hard for this but without neglecting the potentially more important imminent few fixes and QoL tweaks.

To be honest, I have used these similarly related AI things a total of zero times in other applications... so I wonder how useful it can really be (if others can chime in), however, the Depth Map and scene labeling though, that looks interesting.

[ashf]

Could be good but it's not easy task.
Also Affinity still lacks basic features such as non-destructive(vector based) warping so they need to be implemented before A.I. thing...

[kirk23]

I am very excited about everything AI   but yet  have  to see any AI based tool  that could be somewhat helpful when it comes to anything your actual job related.   For some fancy bunny ears in your photos it's super cool for sure.

   What's weird it's surprisingly few really helping things  AI could do so far .  I bought one more year of Photoshop subscribtion last year because of promised AI selecting and photo tagging.    Hardly used it at all.  It's just easier and quicker  to select by yourself  vs  correcting  that stupid  like rock AI .  Tagging  works for anything except what you really need it for.       Same  with sympligon AI lod creation. What a mess.  You end up re-making  by yourself except in some very  obvious  and simple cases.  

Same for AI depth from photo  in  Adobe Alchemist.   It's so-so  really.  Not even  close to real depth captured through photogrammetry.

 

The only good AI job I saw is Octane render de-noiser.    vs other denoisers in other renderers  which basically smear you rendered image into something   unusable and do it only for beauty pass  so also useless

I have mixed impression from AI so far .  A great excitement and expectations and  kind of disappointing reality

[Aaron Martin]

Does serif has any ai implementation plans?

[Revanian]

Bumping this as well. At least a super resolution feature and background/sky removal would seem appropriate. I'm curious too for any plans Serif may have for v2.

[kirk23]

 I vote for Ai trained to do two things.   1 Shadow removing  and de-lighiting .    2 de-noise height maps  from Reality capture

[kyf]

Yeah i have started getting 3rd part programs like Gigapixel AI to fill in the gaps but if Affinity photo falls behind more i will most likely be moving back to adobe.

[billtils]

Why?   What do you actually require as opposed to lust after from the various AI implementations?

[Aaron Martin]

Resizing, removing, generating (vector,pixel or text) would be helpful. Im imagining like if the assistant from current version could have chatting and editing capabilities. I could ask him to generate a vector bird or remove everything green 

 

[tzvi20]

you should not compare adobe to affinity adobe has over 29,000 employees, Serif has less than 100 employees. you should expect that affinity will not have everything that adobe but it has mostly what adobe has but for a much lower price.