PhotoSculpt Textures

[Frenchy Pilou]

Photosculpt Textures

that creates high definition 3D textures or 3D models from 2 photos of the same subject taken from 2 different viewpoints. (20°)

Top Features :

* Very high quality of depth detail
* Very high resolution results
* Very fast : 3d model are displayed on screen within seconds while still processing
* 3d model export (obj, stl)
* texture map export (color, depth, normal, ambient occlusion, specular)
* Tileable texture option

Video
Video
Video
Video

[Frenchy Pilou]

Site Updated : free tilable textures FAQ tutorial screen shoot of the UI etc...

[angelos]

nice. thanks for sharing

[Richard]

Frenchy!!!!

Mate you broken the bank on this one!

Man I've been wanting something like this for ages now. Thank you straight up!!

You're still weird though!

[Lewis Wadsworth]

I put my name on Hippolyte's private 50%-off special offer list. He wrote me back immediately.

I've been fussing about with walls made of irregular masonry and rubble, and this looks like it would be the thing to adequately describe them.

[Richard]

I put through the suggestion that the modeller has option for respecting alpha channel of any imported image to omit construction of the mesh in those areas.

This looks like a very exciting software! I've got some cool ideas going on in my head now!

[hipe-0]

Hello everyone!

Do you accept me to join the conversation too? I can give you answers right away.

My name is Hippolyte from www.photosculpt.net, I'm 34, French too and I designed this tool.

Pilou : Hi! thanks for opening this thread here too. You're unbelieavable, you're everywhere. Next time i'll google your name to save time, not "photosculpt".

Richard: import an alpha to ommit construction of the mesh. That's a very good idea . It might take some time to do though as it need to rewrite key code. I'll think about it. It can be a nice addition to Photosculpt, thanks for sharing. Who's sharing this point of view?

Lewis: "irregular masonry and rubble" You're soo right. I love this kind of subject too from a texture point of view, looking for the right lighting . The only difference is that now I always shoot textures twice to have the 3d model.

Please do not hesitate to ask if any questions?
--
Hippolyte from www.photosculpt.net

[Lewis Wadsworth]

Hello everyone!

Do you accept me to join the conversation too? I can give you answers right away.

My name is Hippolyte from www.photosculpt.net, I'm 34, French too and I designed this tool.

Pilou : Hi! thanks for opening this thread here too. You're unbelieavable, you're everywhere. Next time i'll google your name to save time, not "photosculpt".

Richard: import an alpha to ommit construction of the mesh. That's a very good idea . It might take some time to do though as it need to rewrite key code. I'll think about it. It can be a nice addition to Photosculpt, thanks for sharing. Who's sharing this point of view?

Lewis: "irregular masonry and rubble" You're soo right. I love this kind of subject too from a texture point of view, looking for the right lighting . The only difference is that now I always shoot textures twice to have the 3d model.

Please do not hesitate to ask if any questions?
--
Hippolyte from www.photosculpt.net

Welcome to PPB2, Hippolyte. It's so nice to have a software developer actually ask us what we would like before presenting us with the finished product.

I am not sure I'm quite following Richard's suggestion.

For me, the premier issue is ensuring that any models created with Photosculpt are importable into other software, using the neutral file formats as an exchange medium: not just OBJ, but Collada and DXF as well. Since certain applications prefer lower-polygon counts than others (everyone here should know which application I am describing), it would be particularly nice if Photosculpt would permit one to reduce the face count if necessary (obviously at the cost of some detail) prior to export. Rhino 3D has a preview system for exporting models in mesh format, which allows one to visualize how manipulating the polygon count will affect the mesh prior to export -- it would be nice to see some sort of similar arrangement in your software.

[Frenchy Pilou]

@Hipe : It's a pleasure to make some buzz for very cool progs

[Richard]

Hippolyte welcome to the forums and actually thanks for dropping in!!

Hey your right about Frenchy, he is EVERYWHERE!! I think soft developers should put him on commission - he is like a roaming one man promotions expert!! Never misses a beat on giving us the eyes up on everything that is new and exciting!!

Mate just a quick aside I have posted you to jump in on the "upon release offer" but it keeps bouncing back!

@ Lewis

I am not sure I'm quite following Richard's suggestion.

Mate my suggestion is that if one were to edit out areas of images to be imported to PhotoSculpt (making them alpha channel) - photosculpt would recognise and not infill geometry in these areas. I'd imagine this could be user controlled "do / dont", and several reasons for my suggestion though mainly so areas such as sky where clean up of the mesh later may be more difficult as it is taken into the depth determination of the resulting mesh! Also if one looks at some of Hippolyte's example's of facades where door ways etc produce a rough indent to door ways etc these could be possibly finished clean and the user construct those areas later with more precision.

I think there would be quite a significant number of situations where this may be warranted even if the above examples don't do my suggestion justice!

I think another option either additional or alternatively might be for the inclusion of a sample tool that allows the user to select areas of colour and photosulpt knows that areas with this colour (say 5 pixels x 5 pixels tolerance - the colour being that which the user edits into the photos pre import) are either omitted from the mesh (user defined) or pushed back in the mesh to a defined percentage depth of the overall face solution (again user defined ie 500%). This would allow as exampled (sky / doorways) to be premarked and those areas be push back for easy editing / mesh clean up later. In fact possibly the better option to all this may be the inclusion of a third image, a masking image where the user can say on a white image define areas of say blue as pushing back and red being omitted. This may still allow for the textures to be created correctly yet the geometry manipulated!! In the door case for example the geometry could be pushed back deeper and than later a simple plan be put in and the user paint the facade texture to this ending up with a nice clean surface where needed!

I'd suggest this later option may assit Hippolyte in his coding efforts as it wouldn't rely on recognising the alpha channel at all!

BTW I certainly agree with Lewis on the ability to poly reduce the result!

[hipe-0]

Welcome to PPB2, Hippolyte. It's so nice to have a software developer actually ask us what we would like before presenting us with the finished product.

I am not sure I'm quite following Richard's suggestion.

For me, the premier issue is ensuring that any models created with Photosculpt are importable into other software, using the neutral file formats as an exchange medium: not just OBJ, but Collada and DXF as well. Since certain applications prefer lower-polygon counts than others (everyone here should know which application I am describing), it would be particularly nice if Photosculpt would permit one to reduce the face count if necessary (obviously at the cost of some detail) prior to export. Rhino 3D has a preview system for exporting models in mesh format, which allows one to visualize how manipulating the polygon count will affect the mesh prior to export -- it would be nice to see some sort of similar arrangement in your software.

Thanks! I'm actually really happy to be with you discussing about that.
Those are great ideas you're saying.
Photosculpt is a new born software, it doen't have those extra functions yet, that's just the beginning, please be patient. I see the obj and stl export as the base as every software can read one or the other. Now we can add more stuff.

We're going to work on new functions on an "urgency" basis , so I'm really interrested to hear what are your most interested to have first.

What are the advantages of collada and dxf (over obj)? I don't use that, can you explain?

here http://www.tdt3d.be/articles_viewer.php?art_id=99
and
here http://wiki.cgsociety.org/index.php/...on_of_3d_tools

It says that collada was not read by cinema4d, lightwave, 3d coad and zbrush, is that a fact?

Reduce poly count: That's a great idea! But that's not really easy. Thanks for the interface screen dump that helps me understand.

To your sentence "everyone here should know which application I am describing", well sorry I don't know them all :(

[hipe-0]

Hippolyte welcome to the forums and actually thanks for dropping in!!

Hey your right about Frenchy, he is EVERYWHERE!! I think soft developers should put him on commission - he is like a roaming one man promotions expert!! Never misses a beat on giving us the eyes up on everything that is new and exciting!!

Mate just a quick aside I have posted you to jump in on the "upon release offer" but it keeps bouncing back!

@ Lewis



Mate my suggestion is that if one were to edit out areas of images to be imported to PhotoSculpt (making them alpha channel) - photosculpt would recognise and not infill geometry in these areas. I'd imagine this could be user controlled "do / dont", and several reasons for my suggestion though mainly so areas such as sky where clean up of the mesh later may be more difficult as it is taken into the depth determination of the resulting mesh! Also if one looks at some of Hippolyte's example's of facades where door ways etc produce a rough indent to door ways etc these could be possibly finished clean and the user construct those areas later with more precision.

I think there would be quite a significant number of situations where this may be warranted even if the above examples don't do my suggestion justice!

I think another option either additional or alternatively might be for the inclusion of a sample tool that allows the user to select areas of colour and photosulpt knows that areas with this colour (say 5 pixels x 5 pixels tolerance - the colour being that which the user edits into the photos pre import) are either omitted from the mesh (user defined) or pushed back in the mesh to a defined percentage depth of the overall face solution (again user defined ie 500%). This would allow as exampled (sky / doorways) to be premarked and those areas be push back for easy editing / mesh clean up later. In fact possibly the better option to all this may be the inclusion of a third image, a masking image where the user can say on a white image define areas of say blue as pushing back and red being omitted. This may still allow for the textures to be created correctly yet the geometry manipulated!! In the door case for example the geometry could be pushed back deeper and than later a simple plan be put in and the user paint the facade texture to this ending up with a nice clean surface where needed!

I'd suggest this later option may assit Hippolyte in his coding efforts as it wouldn't rely on recognising the alpha channel at all!

BTW I certainly agree with Lewis on the ability to poly reduce the result!

I'm sorry to hear that my email is not working for you, damn, can you send the bouncing error? :(

user imported Alpha channel to recognise and not put geometry : that an idea! it sound possible and even not overly complicated on a programming side.

Recognising alpha channel automatically : that's not working at all you're right. Too uncertain.

Using color key : will work only on skys and mono color background. I prefer the first proposal as it is way more flexible. What do you think?

I had difficulties to read you because of the language, I'm hoping I've answered to all your questions?

[Richard]

I'm sorry to hear that my email is not working for you, damn, can you send the bouncing error? :(

user imported Alpha channel to recognise and not put geometry : that an idea! it sound possible and even not overly complicated on a programming side.

Recognising alpha channel automatically : that's not working at all you're right. Too uncertain.

Using color key : will work only on skys and mono color background. I prefer the first proposal as it is way more flexible. What do you think?

I had difficulties to read you because of the language, I'm hoping I've answered to all your questions?

Mate what I was meaning about using a ky colour is similar to that suggested in regards to the user created apha map. The user would put in three maps!

1. Left image as per current input!
2. Right image as per current input!
3. Colour reference image created from left or right shot that is used as a reference to the corresponding left or right shot (user defined) - this map could use three colours:

white (255,255,255 / 10 tolerance) - area of geometry to be created
black (0,0,0 / 10 tolerance) - area of geometry to be omitted (alpha)
blue (255,255,0? / 10 tolerance) - area of geometry to be pushed back - user defined percentage of depth!

I'm suggesting the 10% tolerance due to Photoshop's default colour profile as not permitting 100% black - this has always been an issue in creating accurate depth maps (displacement) and therefore would avoid the user asking the question of why funny results are achieved!

I would imagine workflow would be that the user input the three maps and then use a dialogue window in photosculpt that permits an eyedropper selection of the alpha colour (black) and the color representing the colour to be pushed pack (blue in this case).

In regards to Lewis's suggestion of reduced poly option -

I note from an image on your web page that the geometry is progressively developed from - rough low poly through to detailed high poly!

Could there be a user defined option to create a temp file of the mesh at each progressive or every second / third etc and flick back through each option in the preview window to decide which they wish to save as the final?

This would achieve the option Lewis (and many more) would want for whilst allowing you to achieve the same result as the requested poly reduction with limited variation to the current script!

By the way the application that lewis is refering as not handling higher poly counts is Google SketchUp - our tool of choice on the PPB!!!

[Lewis Wadsworth]

Thanks! I'm actually really happy to be with you discussing about that.
Those are great ideas you're saying.
Photosculpt is a new born software, it doen't have those extra functions yet, that's just the beginning, please be patient. I see the obj and stl export as the base as every software can read one or the other. Now we can add more stuff.

We're going to work on new functions on an "urgency" basis , so I'm really interrested to hear what are your most interested to have first.

What are the advantages of collada and dxf (over obj)? I don't use that, can you explain?

here http://www.tdt3d.be/articles_viewer.php?art_id=99
and
here http://wiki.cgsociety.org/index.php/...on_of_3d_tools

It says that collada was not read by cinema4d, lightwave, 3d coad and zbrush, is that a fact?

Reduce poly count: That's a great idea! But that's not really easy. Thanks for the interface screen dump that helps me understand.

To your sentence "everyone here should know which application I am describing", well sorry I don't know them all :(

Hippolyte, I work in architecture, and most of software I use is aimed at that profession. Most of the time, my projects involve Google SketchUp, McNeel Rhino, and occassionally a CAD program or a 3D animation/modeler like Blender or 3DS Max. Whereas the heavy-duty modeling and animation packages, and even Rhino, do a decent job importing OBJ format, SketchUp does not import OBJ. DXF (especially older versions) tends to be a more reliable format for geometry import into CAD programs (and SketchUp, most of the time), although of course it doesn't hold material information like OBJ.

I too have my doubts about the future of the Collada format, but Google has been pushing it as a generic model format for exchanging information (including geometry and materials) between SketchUp, Google Earth, and other programs (here's the list I have: https://collada.org/mediawiki/index....ucts_directory ). I trust that they probably have a good reason for that. At the very least, as a royalty-free open standard the format is not hostage to the whims and profit-incentive of certain companies which periodically use arbitrary changes to their proprietary file formats to extort software upgrades from licensees and impede other software manufacturers from legally producing compatible products.

[hipe-0]

Mate what I was meaning about using a ky colour is similar to that suggested in regards to the user created apha map. The user would put in three maps!

1. Left image as per current input!
2. Right image as per current input!
3. Colour reference image created from left or right shot that is used as a reference to the corresponding left or right shot (user defined) - this map could use three colours:

white (255,255,255 / 10 tolerance) - area of geometry to be created
black (0,0,0 / 10 tolerance) - area of geometry to be omitted (alpha)
blue (255,255,0? / 10 tolerance) - area of geometry to be pushed back - user defined percentage of depth!

I'm suggesting the 10% tolerance due to Photoshop's default colour profile as not permitting 100% black - this has always been an issue in creating accurate depth maps (displacement) and therefore would avoid the user asking the question of why funny results are achieved!

I would imagine workflow would be that the user input the three maps and then use a dialogue window in photosculpt that permits an eyedropper selection of the alpha colour (black) and the color representing the colour to be pushed pack (blue in this case).

In regards to Lewis's suggestion of reduced poly option -

I note from an image on your web page that the geometry is progressively developed from - rough low poly through to detailed high poly!

Could there be a user defined option to create a temp file of the mesh at each progressive or every second / third etc and flick back through each option in the preview window to decide which they wish to save as the final?

This would achieve the option Lewis (and many more) would want for whilst allowing you to achieve the same result as the requested poly reduction with limited variation to the current script!

By the way the application that lewis is refering as not handling higher poly counts is Google SketchUp - our tool of choice on the PPB!!!

Alpha option:
I see what you mean with white/black/blue. That makes sense, here is my suggestion: how about just white/black only? I suggest that pushing geometry back could be obtained with another function. Advantage: more simple for CG people as they already know what an alpha map is.
I'm going to think about it as it changes my directx functions, I cannot do that very quickly. It will have a memory impact too. Thanks for your time!

Other suggestion: What about an alpha threshold at 50% instead of 10%? Discarding all transparency would be more efficient. As polygon can basically be either there or not there. Just a thought...

Poly reduce option:
1. saving different map size: This is done already. Sorry I may have misunderstood you guys from the beginning :rolleyes:(see 2.). Today you can save all different model sizes of your choice using the save multiple option. All in one operation. (This was requested by my friend Jérôme)
You can also browse the subdivisions in one click (like zbrush, like blender). Go up, go down. You get immediate feed back on screen and you can save the one you see doing "save as".
Subdividing has one disadvantage: reducing subdivision means reducing detail.

2. grid optimisation: I initially thought that you meant : reducing polycount by merging/collapsing/grouping polygons together based on shape. I like this idea, but this is hard to do. Who's interested into this?

[hipe-0]

Hippolyte, I work in architecture, and most of software I use is aimed at that profession. Most of the time, my projects involve Google SketchUp, McNeel Rhino, and occassionally a CAD program or a 3D animation/modeler like Blender or 3DS Max. Whereas the heavy-duty modeling and animation packages, and even Rhino, do a decent job importing OBJ format, SketchUp does not import OBJ. DXF (especially older versions) tends to be a more reliable format for geometry import into CAD programs (and SketchUp, most of the time), although of course it doesn't hold material information like OBJ.

I too have my doubts about the future of the Collada format, but Google has been pushing it as a generic model format for exchanging information (including geometry and materials) between SketchUp, Google Earth, and other programs (here's the list I have: https://collada.org/mediawiki/index....ucts_directory ). I trust that they probably have a good reason for that. At the very least, as a royalty-free open standard the format is not hostage to the whims and profit-incentive of certain companies which periodically use arbitrary changes to their proprietary file formats to extort software upgrades from licensees and impede other software manufacturers from legally producing compatible products.

collada + dxf export:
Thanks for your time replying Lewis. It helps me understand why you need them and it makes sense to me. I will think about it for future releases.
I have taken your request into consideration and will study that.

polycount limit on softwares:
I love sketchup. But I know it to be not good for high poly display :(, I tested it 2 years ago on a very basic low poly plane model once. This was really slow already.
I don't know the other softwares are they good on this feature ?

[nightzombie]

wow! it looks amazing. I cant wait to try it out with my modo work flow

[Frenchy Pilou]

Last Video of the interface in action
Video

[Frenchy Pilou]

Photosculpt Textures is realised!

[hipe-0]

Yes PHOTOSCULPT TEXTURES V1.0 is released now !

See: www.photosculpt.net

The base license is EUR 99.00

I'm really happy to finally spread this toy to the community.

Alpha map:
By the way the new function for the alpha map that you guys requested has been implemented and it works great! Testers are really happy with it too. I already have plans to improve it. Thanks for taking the time to discuss with me

See here for more http://www.youtube.com/watch?v=G8wwrInbnfM