3D Reconstruction Beta - A Review
If you're new here, you may want to subscribe to my RSS feed. Thanks for visiting!
Yesterday, June 7th, the developers released a new version of the EON Human service. They are calling it the “Generic 3D Reconstruction Beta.” There are several things that have been revised, updated or changed to improve the quality of the service since last week’s version. It is great to see the service evolve and improve so fast! I will make note of the changes below:
The header banner to access the service has changed. It now looks like the image above. This image adds more detail (a kind of short introduction) about what the service is, icons of the social networks or blogging platforms that it “purportedly” works with to embed the final 3D widget code, and a “Get Started” button. These are all good improvements, although the service will work with a picture that is not your own as well, as they have shown with the celebrity images that flank the EON Human Ning homepage. You must have an account and be signed into your EON Human Ning profile in order to access this button, as Ning will pass your Ning ID to the EONReality.com servers. The URL this banner links to is http://www.eonreality.com/eonhuman/index.php?ningid=************* where the *’s are your Ning ID.
Once you are into the service, you are presented with a new upload page. This new page shows you the specifications of images that work the best with the service:
- Passport Style Photo
- Full Head, Facing Camera or Left/Right Shot
- Mouth Closed
- Remove Hair from Face
- No Glasses are allowed
- Image upload size is limited to 400k.
They also show you visual examples of the types of face photos that are acceptable. They can now accept photographs of 3/4 left and right face photos in additional to straight frontal shots.
Once you have uploaded your image, you are asked to choose which type of face photograph you’ve uploaded - left shot, front shot, or right shot. Frontal shots still tend to give the best quality 3D reconstruction. Pick the shot that is closest to your image.
Then the pick points process starts, which is similar to the first technical release of the service. The difference is that they give you a visual example of the point you should pick, instead of just written instructions. This is very helpful. You are also allowed to choose more points, about 8 instead of 5 like last week, in order to get a better quality 3D reconstruction in the end. The pick points help match the 3D average face model to the orientation of your face photograph.There are a few points that are still a little confusing - “Click on the right side of your jaw” and “Click on the left side of your jaw.” Except for the visual example, it is unclear where on the jaw you should click. It appears that you should put this point on the jaw at the point where the neck meets the jaw, but I’ve found that this does not always appear to be the best solution, and placing it a little lower on the jaw sometimes leads to better results. The point “Click on your chin” is also vague. Where on my chin? Except for the visual example given, it is hard to determine where to click. It appears as though it should be at the furthest out tip of the chin bone.
Once you do this and click on “Generate Face” you are taken to a new popup window (why a popup?) of the process box that reads “Generating 3D Face - This may take a minute… Please do not refresh your browser. Meanwhile Enjoy These Movies.” You are then presented with some movies on EON Human, the same ones that are on YouTube for your enjoyment and entertainment while the server processes your image and generates the 3D model.
Once it is complete you see “FACE GENERATION COMPLETE!! CLICK HE.” It probably says “click here” but the words run off the button.
You are then presented with your new 3D Face Reconstruction. You can rotate it left and right with the buttons. It would be better if they gave you a real-time interactive slider to rotate the face around, with instantaneous refreshes of the image, almost like a Quicktime VR type of interaction with your model. At any rate, if you are lucky you will have a pretty good representation of your face in 3D. You are then allowed to “Post It” or “Restart.”
Clicking “Post It” takes you to the 3D carousel page where you can see the differenc choices for face manipulation - 3D face, Feminine, Masculine, Fat, Thin, Beauty, Ugly. You are also given the option to “View Original.” Two different embed codes are shown at this point - “Social networks, Blogs & Websites, Copy Code” and “Video Sharing Upload your 3D Spa Face Video to EON Human.” I am unsure what the difference is between these two code sets because they appear to give you identical code. They both show the 3D Reconstruction widget, which looks like this (continue below):
The “View Original” button works now (although we don’t need the slider), but the “Tell a Friend” and “Make Me” still do not work. The images on the carousel also appear larger.
This new version of EON Human looks like a good step forward in improving the service. The 3D model that is produced is more accurate because of the increased number of pick points, and you can now use left and right face photographs in addition to frontals. Here are some of the things that I would like to see improved:
- A real-time interactive slider that allows rotation of the 3D face in real-time as you drag it right or left, perhaps even being able to drag the model around up and down also. More of a 3D interaction…
- The ability to choose which 3D Spa treatments you want in your embedded widget. What if I only want the 3D Face, and not the Fat, Thin, Beauty, Feminine or Masculine?
- Integration of the 3D model back into the original photograph.
- The ability to adjust the various 3D Spa treatments at any rotation point of the face, not just the front.
- Being able to rotate the 3D model to a full profile view on the left and right sides, not just 3/4 view.
- Adding the capability to interactively adjust certain parameters of the embedded widget code for height, width, color, face manipulation features, etc.
- Better instructions on how to embed the code in a social network (it is very complicated to get it into Facebook), blogs, or websites.
- Giving me an “account” where I can go back to the models I’ve made in the past and adjust them or get their embed code again.
These are just some of the things off top of my head. I’m sure I’ll think of more in the coming days. But congratulations to the EON Human developers for continuing to produce a new, fun, and exciting service that will no doubt turn out to become very popular and useful, not only for entertainment, but in other areas of industry as well.
Comments
5 Responses to “3D Reconstruction Beta - A Review”
Leave a Reply
Thank you for a VERY good Review.
We will take your suggestions and continue to improve this service.
Dan
can we get an .obj of the output?
joe
OBJ output is not yet a feature of EON Human, but may be offered in the future.
Is there some other 3d output format? Otherwise, although it is a cool demo of what must be a very complex process, what is its benefit to humanity and the future and the ubermenchen?
thanks
joe
There is no 3D output in this “consumer” version. They have an enterprise version that would probably export in many different 3D formats, but the cost is tremendous. This entry-level service is not for technical users, but to create representations that can be embedded on websites for entertainment.