Hi all,
Please use this thread as a place to consolidate feature requests for the QCAR SDK. If there is something missing that is holding back your development efforts, we'd like to know about it! If you're having trouble with an existing feature and would like more sample code to help you get started, we'd like to know about that too.
Looking forward to your input, thanks!
- Kim
I would change my previous wish to something else.
As mentioned below, a way to programmatically manage targets from an API.
This way we can create multiple databases and load them on our server where they get then downloaded on demand.
This would make managing multiple database apps a lot easier.
The target manager needs a serious update and more flexibility
it will be good if you could provide a ready to use prefab for User Defined Target Builder so i don't need to invent the wheel and just implement it inside my projects
right now i struggle to create prefabs from it . it say that i can't.
can you make it as Packets That can be inserted only with what needed . you have all the scenes in one project it will be good if you divided them
because sometimes you work on a project with other scripts that can collide and destroy the project (early version with old ones) and so on issues.
thank you
yodda
For the website:
When people upload their images and get low scores it would be nice to see a reason or suggestion for improvement for the low scoring items, so people can create better items. I do know you have documentation for this type of thing but it would be nice to see more dynamic input based on specific images.
Hello,
Instead of loading all the targets from a database and activating them (which requires a lot of time when initializing vuforia and creating the datasets)
Would it be possible to activate just the ones we needed?The only solution at the moment is to split databases into multiple ones, by doing so it would become a mess very quickly.
You are correct that there is no feature to load individual targets from a single device database. I will escalate your request to the product team for consideration.
Also, I opened up the database with 7zip and noticed that also grayscale images are imported into the database, are these needed for vuforia to work correctly?
Yes. All assets in the DAT file are required for a Vuforia Engine device database to function properly.
Thanks,
Vuforia Engine Support
Instead of loading all the targets from a database and activating them (which requires a lot of time when initializing vuforia and creating the datasets)
Would it be possible to activate just the ones we needed?
The only solution at the moment is to split databases into multiple ones, by doing so it would become a mess very quickly.
Also, I opened up the database with 7zip and noticed that also grayscale images are imported into the database, are these needed for vuforia to work correctly?
Hello @lwarner,
I would like to see the return of UWP builds in the newer versions of Vuforia. Even if they support is not provided in case of an issue, but It would be nice to see the continued support. Same thing with Standalone builds.
Can you be more specific about what artifacts are missing now that Vuforia Engine has provided in the past?
Thanks,
Vuforia Engine Support
Feature Request
Access device database via API.
Why?
To manage targets and download device databases programmatically.
---
Another great way would be to open source the device database creation source code so that developers can create device databases on their own.
Hello @CeciLisCosta,
Have you reviewed our External Camera feature? https://library.vuforia.com/content/vuforia-library/en/articles/Solution/external-camera.html
Thanks,
Vuforia Engine Support
PROBLEM:
I urge the Vuforia development team to increase the robustness of Vuforia API. It's the Vuforia API that convinced me to use your product. But like all frameworks, I immediately bumped its limits.
For example, the Vuforia 3d scanning tool use is limited. My application requires unique feature point extraction when building a dataset for a 3d object target. Such that I want to build my own custom 3D scanning tool. In particular, I need five independent, detectable regions of a single 3d object.
I suspect there are many developers that require an extensible AR framework that adapts to every application possibility.
Only a rich API of low level and high level functions can satisfy our needs. If you make the API too simple (e.g. Wikitude, etc.), then it will always remain a toy. And not a professional developer's go to framework.
WISH LIST:
Provide low level functions to manage collections of feature extraction points during target dataset creation.
get_featurePointsList()
remove_featurePoint()
add_featurePoint()
modify_featurePoint()
save_featurePointsList()
Provide feature extraction points that discern color contrast for object pattern recognition.
We live in a colorful world, yet it seems your feature extraction functions rely exclusively on luminosity -- and not hue.
Ken Huebner
Owner / Software Engineer
Huebner Design USA
There is a high demand for combining Ground Plane and Image Targets. It should be able to get the Image Target location in Ground Plane and vice versa.
This would have mulltiple use cases, as discussed in this thread: https://developer.vuforia.com/forum/ground-plane/combining-ground-plane-and-image-targets
20000+ vies already.
Problem
It's hard to manage a large number of targets via the target manager (web interface) without any simple search functionality.
Simplest solution
Add a additional select option to the pagination dropdown list to show all targets (this is limited to 200 at the moment).
This way someone who has to manage large amounts of targets (like me) can at least use the browser search function to find a uniq target without the effort of paginating through N pages.
Better solution
Add a simple search bar (doesn't have to be fuzzy for now) to search for a exactly match of the target name.
e.g. searching for my_target_1 will only show my_target_1
Even better solution
Add a fuzzy search bar to return a collection of everything that could be interesting.
e.g. searching for my_target* will show my_target_1, my_target_2, ...
Very much agree, particularly about the 3D point cloud!
- ARCore publishes its internal point cloud
- ARKit publishes its internal point cloud
- Smart Terrain has been deprecated
So if an application wants to do hit-testing against the real world, Vuforia Fusion + Ground Plane sounds like a good option, but in practice it is not usable (please correct me if I'm wrong). Instead, developers have to revert to the underlying platform-specific libraries.. Please add this as a feature to 7.2!
Publishing the 2D points is also useful: then we can give better feedback to the user about the usability of the current environment for tracking.
I think the delete database function is a fundamental need.
The only reason I can think about for not having this feature is that Vuforia is not actually deleting the targets I've uploaded and you need the databases to index them.
In other words, my suspicious is that instead of removing my pics you're just adding a isDeleted=true field in the DB.
Thanks
I really wish the viewforia forums would stop with the inverse-response style forums. The posts are in reverse-chronological order which just makes no sense! I guess you guys did it so that the answer to the question in the OP would be at the top of the page (assuming the answer would be the final post in the thread) but more often than not the final post is some offhand context-less answer and you need to trawl through the thread anyway just to find the original question was nothing like the on you were asking. You guys could maybe use the stack-overflow model? I'm pretty sure they whitelabel that platform...
Sometimes the Camera Distortion affects the pose estimation result, makes it inaccuracy.
So I wish vuforia can expose the ability to set device camera intrinsic parameters while running on embedded cameras.