Hello Numenus,
if you are looking for some input for your High poly density viewer, I have some for you:
Hi Mattia, thanks a lot for your valuable input! I will comment on your input.
- What do you think to implement a "Section Tool" to get a real time section-plane? Ok, I know that this tool is already present in many 3D viewer, but they always cut the 3D objects on their profile. So, we can see the inner part of the object (like a negative surface) and the sectioned profile. Is possible to get a section-tool with the filling effect? a tool that cut the object in real time with a virtual plane and that show the 3D object as solid object?
I like this idea, and we will implement this feature. Filling the object is an interesting idea and I think we will be able to implement that as well - I already have an idea
Of course this will possible in realtime or interactive frame rate depending on the scene and computer. It won't be slower than usual rendering I guess.
- A poly counter is already present? if no, will be useful a poly/vertex/triangles counter in the main window of your program, easily to consult.
A poly counter we have, adding vertex counter is of course pretty easy and will be done.
- Can you import multiple objects in the same scene? if no, can be useful this feature, with objects like my Suzuky bike, made up of many parts. So, we can select and show/hidden the selected pieces, as we want.
Loading multiple files is already possible. Moving these parts around also is possible, tough it is not comfortable yet, since it is still in development
The ability to show and hide selected pieces will definitively be implemented soon as well.
- You think to implement some other feature, like poly reduction and/or the generation of normal/bump/displacemet/ambient occlusion maps? If yes: at present day don't exist a software that do all this in the same time. I know that exists many expensive software that can do poly reduction and creation of maps from High poly mesh... but don't exist an easy and "all-in-one" software to get a low poly mesh with the application of normal normal/bump/displacemet/ambient occlusion maps directly from High poly mesh. This can be useful for a better workflow of who, like me, want to share online the scan results.
Okay this sounds really interesting as well. Implementation of such features is a bit more work of course. So I would say this will not be included in a first demo version and maybe not in the 1.0 Version - but I definitively keep this in mind and like this to have in a future version of our software. And as you said, I would intend to make it pretty easy to use, tailored to the specific need of scanned models. Currently we supprt bumb mapping and ambient occulusion already, but we don't export these into textures yet.
- You wrote that your software can export videos... so is already present a "turntable" feature? I'm thinking about something that can perform a rotation (with adjustable speed) all around a chosen axis, and automatically save a video of the turntable movement.
Yes, we have a turntable option. I can upload a video later which shows the result. It is not that configurable yet, but of course it's no big deal to make it more flexible. Currently we only export an image series, but we want to make it possible to export an video directly, of course.
- What about the shaders used into your viewer? Standard shader (Blinn, Phong, X-Ray, wireframe,...) are no sufficient to get the best look of our scanned model. If you point at the ZBrush site, you can see great shaders, like wax material, pewter material, old metals materials ... Is possible to have shader cool like they?
We support a couple of shaders so far, however extending these with more and more shaders is no big deal and only work that need to be done
Since we have a good plugin architecture, shaders can also easily be added after a realase and with our SDK it is even possible that people who can programm C++ are able to write their own shaders. Besides we want to extend our shaders as well of course. Maybe the community can compile a list of must-have-shaders
And now some other questions about your project:
- What kind of technology there is behind it?
The core technology is a real time ray tracer called Augenblick, which I developed during my Ph.D. time at the university of Koblenz. It is a highly optimized ray tracer which still is very flexible and easy to extend. My ray tracer is obviously not the only one in the whole wide world, but most of them are either fast and unflexible or they are very flexible but slow. Augenblick is of course multi-threaded, so it can use as many cores as the machine has and supports SSE up to SSE4 (if available) for instruction paralellism. However, through a smart abstraction layer our SIMD ray tracer also runs on PowerPCs (any old macs around? ^^) as well as without SIMD at all - which is of course slower. We don't intend to add network support just now, since usually private users don't want to buy a cluster of machines to get results fast
- How long is the time required to load into the pc memory an high poly mesh? For examples a viewer like GLC require too much time (and some time crash) with model of millions of polygons. ZBrush have an implemented importer tool that allow to load huge 3D file in less of one minute... can you estimate a load-time for the complete model of my Suzuky .obj?
I can give you exact numbers for you model
We have not focused on that, so laoding times are still rather slow. For your 5+ million poly mesh it currently is 1.45 min. But right now we are rewriting the OBJ loader, since our current one is a bit old and can be improved a lot. Actually around 48 seconds is just copying the file into memory and another 48 seconds is parsing it.Only around 5 seconds is Augenblickinternal stuff, like allocating memory and creating acceleration data structures and stuff like that. So I guess after optimizing we will be able to load it clearly below one minute.
About a suggested price... is hard for me to estimate it, without any chance to test/try it. but if we are talking about a "simple" viewer, I think that a price between 25 € and 50 € is good, but if you can implement other stuff (poly reducer, material baking from high to low poly meshes, cool shaders, ...) I think that you ca set the price bar between 100 € and 200 €.
Of course, I understand it's hard to guess. But from my point of view your suggestions sound reasonable. Well, since we are very early in our planning stage I cannot make any promises, but I can say, that I really like the idea of developing an application for the community, since it fits so well into what we want to do with our ray tracer. Please don't take this as an official schedule, it is just some good guess I can make at this point in time: I would like to have a demo version ready in let's say around two month - and then maybe a first version in around 6 month maybe a bit longer. For the demo version I think the best idea is to limit the maximum resolution and maybe add a watermark or something to the images. Otherwise I would keep the application fully functional.
Thanks a lot for your input! If you can think of anything else, don't hesitate to tell us
Best regards,
Oliver
Anyone else have this problem?
.