40 camera headscanner, results

  • 14 Replies
  • 1599 Views
40 camera headscanner, results
« on: 23 / October / 2017, 12:56:34 »
Advertisements
Hello everyone,
 I think I read on this forum that the contributors/developers are a bit disappointed/disheartened sometimes when new people show up on the forum for a few days, get the answers they need, and then disappear and are never seen again.

 So I will post a few pictures/results, so you can see what your time and effort has made possible.

 I'm attaching a screenshot from an early head scan, as a quick preview, and for those who are still interested, also a link to 2 scanned models on Sketchfab, which allows you to view the scan as a proper 3d model, with the abilty to spin it round, zoom in and out, and move side to side:

https://sketchfab.com/ottokar_ps

thank-you to all the developers and contributors,
andrew

*

Online reyalp

  • ******
  • 11447
Re: 40 camera headscanner, results
« Reply #1 on: 23 / October / 2017, 13:14:58 »
So I will post a few pictures/results, so you can see what your time and effort has made possible.
Thank you, seeing the cool things people do is one of the best parts of this project :)

Any details you would like to share about the your setup are welcome.
Don't forget what the H stands for.

Re: 40 camera headscanner, results
« Reply #2 on: 24 / October / 2017, 01:54:06 »
So I will post a few pictures/results, so you can see what your time and effort has made possible.

Very interesting...  :)
I would like to see pictures from configuration of the scanner...
Which cam you are using?

Re: 40 camera headscanner, results
« Reply #3 on: 24 / October / 2017, 08:00:28 »
Great stuff Andrew!!


Re: 40 camera headscanner, results
« Reply #4 on: 30 / October / 2017, 21:21:32 »
Thank-you for your replies and interest. I'm attaching a photo of the setup I used for those scans, I just dismantled it this weekend as I want something more practical and easier to coexist with, in my very small house.

 I think the physical arrangement was unusual (compared to other human scanning rigs I have seen) in having a top beam with cameras attached to try and capture the hair well, though I am still getting rather poor results with the hair.

 The cameras are a mix of types - from A470 and A550, up to Ixus 160 and Sx210. Having  a mix seems ok, the photogrammetry software seems to cope with the differing picture sizes and lens parameters.

 I'm not sure if the quality and sheer pixel resolution of the older cameras is good enough, but it has let me get started at an affordable price.

 The system at the moment is not a practical solution, it's more a case of seeing what quality level I can get with these cameras, and then trying to change the system to be more practical and seeing how much quality I have to give up with each change,

andrew

Re: 40 camera headscanner, results
« Reply #5 on: 01 / November / 2017, 16:37:19 »
Memory is that scanning hair, via photogrammetry, could be difficult until sensor resolution is sufficient for a single hair.

Even then, that is before the difficulties associated with occluded volumes.

Thanks for sharing interesting setup. 

*

Offline mphx

  • ***
  • 208
Re: 40 camera headscanner, results
« Reply #6 on: 19 / November / 2017, 09:26:19 »
Very good result . Few pointers though :


1.Lighting - You need strong lights all around the object/head/human/whatever .If you dont have the correct lighting you end up with shadows on the texture (as i can see in the 3d models in scketchfab).
And then u need an experienced 3d designer to "fix" these shadows by painting texture.Hard job.
2.Mesh - Resulting models were very good , although some smoothing/scultping could be done for even better result :)
3.RC - Half credits go to this program..worth every cent it costs...we created a scale model of a whole (well part) of a city and if we didnt have RC we will still been waiting for photoscan to align the photos (thousands of photos) and the result was way better of any other photogrammetry software.
4.Hair - Your least concern.In our case where we print the model , we dont need accuracy in hair.We just get the general form of hair , we sculpt it a bit , cut any hair waving in the air if any , and done.
Of course some 3d designing skills are required.
« Last Edit: 19 / November / 2017, 09:29:07 by mphx »

Re: 40 camera headscanner, results
« Reply #7 on: 23 / November / 2017, 19:24:51 »
thanks Andy,

Mphx - thank-you for your detailed advice.

 That very long thread you started on your business was very informative, I would be very keen to hear how you are doing since then.

 Could it be that the technical side of things has not changed since you got a working setup, and that it is only the business side of things brings changes and challenges?

 Even just on the business side, I'm sure everyone on this forum would be interested, things like how this new idea is catching on with the public, given the difficult economic situation in Greece, what is the demand for this non-essential service, do people's expectation match up to reality, do you have many repeat customers or do people tend to just buy one 3d print and so on,
thanks,
andrew


*

Offline mphx

  • ***
  • 208
Re: 40 camera headscanner, results
« Reply #8 on: 25 / November / 2017, 20:50:45 »
@ottokar


Technical side of things has not changed alot.At some point we added 3 projectors projecting a pattern on the human being shot , so RC can find more reference points and align photos more easily.
RC has no trouble aligning photos of big distances ( e.g. drone photos) but when it comes to camera rig where the object/human , being shot is relatively close to cameras ,it has hard time (some of the times) to get a correct align.So we help it with the projected pattern :)
Other than that , setup is working as intended , with minor problems mostly because usb connections get loose some times and cameras lose connection .Nothing serious.


The other part of the whole project is the business part.
When we started it like 2-3 years ago , it was a new idea for Greece , and there was a hype..tv channels wanted to cover the whole thing , we made some live tv appearances and some recorded ones :) Sites wrote some articles about us , people were coming in numbers.Celebrities wanting to cash in their "x followers on y social media"-advertise to get a free miniature...Good times.
Then we had "capital controls" applied in our economy system.Practically people couldnt withdraw more than around 800 euros per week.So as you said , 3d printed miniatures was and still is a non-essential service and given the restriction on money withdraw , people would rather spend these weekly money on survival things rather than 3d printed stuff.
So there was a reduction on people coming to get their face scanned :) But still people kept coming ..
But as you suspected we have the "repeated customers".Usually parents that scan their children at periodical intervals to get their children's growth on miniatures.We have many of them.
Others that liked the first one and get their relatives/friends to get one.And of course there some that got one and that's it.
Business is low at this time..but at "good times" , the whole rig paid out its money , so now it is working for the "extra coin" since we do 3d modelling/designing/animation works at the same time that pay more and in less time (clear profit).


Do people's expectation match up to reality?This is a good question.The answer can't be yes or no.It is yes and no.Its related to someone's personality.
Keep in mind tha people dont usually see themselves in 3d perspective in their real life..they can't see their back from outside or their sides.So , many people when they see their miniature are a bit confused ...its their first time they see themselves from all angles.These kind of people have some minor complaints like "my butt isnt like this" , " my ears arent that big" , etc etc.
We explain them what is happening (to them) , they get it and its all good.
Then we have the other category of people.Those that aren't satisfied with anything..either its about miniature or anything else in their life.No matter how perfect you make their miniature..they refuse to see that..they look like this.They live in denial in general :)
There are others that they think that because they paid a shitload of money they must complain about something..to feel that they are in control of the whole thing.
We have the major category "women" , that they want a cheap plastic surgery (in miniature) , because probably the real one is too expensive :P


Seriously now , complains from all of the above categories are rare..like 1 out of 20 miniatures..most of them are jumping up and down when they get their miniatures out of happiness :)


PS : Sorry that took me 3 days to answer/write something..mail notification went straight to junk mail..wtf...
« Last Edit: 25 / November / 2017, 20:55:07 by mphx »

Re: 40 camera headscanner, results
« Reply #9 on: 05 / December / 2017, 18:08:26 »
Hi Mphx,
 thank-you for such a long, informative answer.

 Your use of projectors I find *very* interesting. I've downloaded the sample photos from the pi3dscan website, and I was amazed at the quality of the models from them, presumably due to their use of projectors and despite using inferior Raspberry Pi cameras.

 I have lots of questions:

 1) Did you buy projectors because you were dissatisfied with the quality from the Canon cameras on their own?

 2) I imagine the quality with the projectors is good enough that you don't need to buy more cameras, nor do you need to buy better quality cameras (like DSLRs) - is that right?

 3) Are your 3 projectors enough to cover the whole model, or would you like to have more?

 4) How do you do the timing between the plain shot and the shot with noise projection? With Pi3dscan, they do a normal photo with all cameras, then the projected photo with all cameras, and there must be some kind of delay between the 2 shots, perhaps quite long if a photo has to be written to the sd card. If i were to do it (which I am considering), then I would perhaps have 50 cameras, split between 15 just for plain photos, and 35 for photos with projection, so that I don't have to wait excessively long inbetween.
 Those photosets available to download on the Pi3dscan site - I don't know if they're typical results, or if some (perhaps only a small minority of) models are especially good/lucky at not moving between photos, and only those photosets were made available for download.


 Finally, another question on the business side, partly brought on by one of your questions in your own thread: you've built up this business, developed the technical side to get it all working, and so on. Now, am I right to think that you can't really franchise your know-how? If you help someone start a similar business in another city, you could charge them for your help, but once they have a working setup, they wouldn't feel the need to pay you an ongoing fee, they could just stop paying you and carry on using the hardware and the scripts that you've set them up with.

many thanks,
Andrew

 

Related Topics