If you want a flavor of how good AI can be these days – check out our new 2D3D image conversion site LeiaPix Converter. Add depth to ANY picture, download and share as GIF, MP4 or our own Lightfield Image Format (LIF) to view on Red Hydrogen or Lume Pad.
Our gift to the community, it’s FREE, enjoy !
Awesome! Just curious If the same algorythms are used as in the Standalone APPS on the LumePad, or if the Training data or Depth estimation is any different.
Nice.
Saving a file from LeiaPix Converter as LIV gives me a .jpg file. Is there metadata, or do I need to save to a particular name format?
@Johannes_Hainer The version used in LeiaPix Converter is GPU-accelerated in the cloud, as opposed to being done locally on a mobile SOC when you convert on Lume Pad.
@MarkJeffcock LIF files are in the .jpg file format, the metadata is already encoded into the image and will be readable by Lume Pad in LeiaPlayer. Just make sure not to use any methods of transfer that might compress the image! For example, don’t use Facebook Messenger to send the image from LeiaPix Converter from your PC to your Lume Pad.
@Johannes_Hainer the web version uses a much more beefed up neural network the results should be significantly better than on the mobile version. In time we will let you guys access the cloud service directly from the mobile app…
Cool, thanks for the Reply, also to @Nima ! You Guys are Awesome, these are definetly exciting Times for anyone interested in 3d and Lightfield Technology.
I hope that all the Covid Tech shortages and crumbling markets aren’t hitting Your Business that mouch, and that You can keep pushing Your Ideas and Visions for many Years to come.
As a Long Time 3d enthusiast that constantly searched for Innovation and new findings it made my Heart jump to See this kind of Technology finally come to Life.
I have read a Lot of papers over the last decade and most of the concepts where pretty amazing in their specific field, but seeing it all come together is truly Magic.
nice! one of my favorite things about ML/AI is that the results should only get better over time.
great work! was very stoked to see this. can’t wait to play around with the tool some more 
LOVE the inclusion of the depth map editor
yes and careful the depth map editing is super addictive 
I have accessed successfully via PC, but apparently the website is labeled as dangerous on the mobile device (see the screenshot)
Does the alert go away when you use https instead?
Good point before writing that statement I should have tried. The answer is it doesn’t, so the problem is somethings else.
@m.farina seems to be an issue with McAfee smart filtering they have been contacted to try resolve the issue…
I finally got a chance to test it out on a very complex photo. I’m very impressed. I posted the photo (nighttime polo on bike) on leiapix under my name.
nice one @PeteSeattle !
Perfect: issue solved. Also immediately tested with some picture on my H1, and results are really great!
super !
What is the procedure for making feature requests for the LeiaPix Converter? I have several requests, is it better to put them all in one post, or some other way?
Thank you. I first want to say that I like a lot of features of the present converter. But the depth map editor could be improved.
I don’t know the true smallest size of the editing brush in pixels, but I find it doesn’t get small enough for detailed repairs, e.g., the gap between a person’s legs when they are standing.
there should be a visual indication of the size of the editing brush. In Photoshop, the cursor is replaced with a white circle whose diameter is the brush size. (Right now, to judge the size, I have to set hardness and opacity to max dab with the brush, observe the size of the circle it makes, then undo. That is incredibly kludgy).
again for detail work – hair and fur for example – I would like to be able to zoom in on part of the image.
you provide an opacity control for the brush. But I would also like an opacity control for the layers. That is, in edit depth map mode there is a depth map (grayscale) overlaying the source image (color),. But sometimes, to see where to make the change, we need to see the source image better, which means making the depth map more transparent.
related to that, it would be nice to be able to make selections based on either the source image or the depth map, a la magic wand tool in Photoshop, and then limit the effect of the brush to the selected area.
right now, editing is absolute – set level explicitly or with an eyedropper, and paint with that. But it may be advantageous sometimes to employ relative editing, that the brush lightens or darkens the depth map beneath it, to an extent determined by the opacity and possibly some max increment/decrement setting. This is roughly equivalent to dodge (lighten) and burn (darken) in darkroom photography.
there may be limits to what can be done with the internal depth map editor. So it would be nice to save the current depth map as a separate file so it can be edited in another graphics program. Which means it should also be possible to load a depth map and associate it with a source image.
(With external edit capability provided, some of the other suggestions become less critical.)
I understand from the developer doc that the LIF file format include the depth (disparity) map info.
