Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - Tyler J

Pages: [1] 2 3

I did the same thing, those camera are always on auto mode, meaning you will get 25 pictures with different ISO, and aperture speed.

Photogrammetry reconstruction needs pictures with the same ISO, aperture speed, and aperture opening to generate good result.

I have a 25 Coolpix array and almost all the times 5-10 of them can't focus properly.

With only 10 second hand canon 600D (DSLR) you'll get much better result.
From my experiment I saw that if all the camera focused properly then only 10 camera were required to get a face scan.

What about pricing? could you services could be use by a student to clean some face scan for example?

what is processing time to reasonable time-frames?
1? 2h? 12h?

just go for a current 4 core 3.5Ghz I7, a ATI R9 290 and with 10 pictures you'll get processing time of around 2h with parameters set to high.

For triggering camera with no latency I use cables, however all cables must be of the same lenght to insure no delay betwen each of them.

So I got a 600D and a 1100D, there does not seems to be a huge difference.
The 600D cost 2 times more than the 1100D on the second hand market

Comparaison, 600D on the left, 1100D on the right :


F8, TV 1/10, ISO 100, stock 18-55 lens@18mm, tripod, 2s delay, miror locked

One solution would be to use a photogrammetry software used specificaly for human (I don't know any)
If the software knew what a human is suposed to look like and use that (nose, mouth, eyes, ears, chin, etc) to match the pictures then it would be much easier.

Face and Body Scanning / Canon 600D/1100D or other?
« on: May 27, 2015, 12:28:04 PM »

I want to experiment with DSLR.
I can get a 1100D for 220€ or a 600D for 350€ both with a 15-58 lens.

Should I invest in the 600D or the mesh/texture gain would be very little?
Do you know a better DSLR (500€ naked max).


Face and Body Scanning / Best depth filtering for human heads?
« on: April 28, 2015, 12:37:01 AM »

I'm experimenting with depth filtering, however trying various combination will take 2 weeks.
I'm working with 21 picture, only the face.

Is there a depth filtering that always suits human heads? or is it pictures quality dependant?

Experimenting with :

no alignment optimisation
   no removal of bad sparse point
   removal of bad sparse point
      medium dense cloud
      high dense cloud
      very high dense cloud
         agressive depth filtering
         mild depth filtering
         moderate depth filtering
         depth filtering disabled

Done so far
no alignment optimisation
removal of bad sparse point
very high agressive dense cloud
high mesh

comparaison will come

each step take me one day or more



I did this with point and shoot camera, be sure to take some CHDK enabled canon camera.
Otherwhise each camera will choose it's own settings (ISO, aperture speed) but you need the same
settings for each pictures.

Also you can use CHDK to take picture via USB.

Since mine where not CHDK enabled my scans are not always good and I had to solder wire on the camera board :/

For alimentation you can get some nice DC/DC converter on ebay for 1$ each and use a computer 12V alimentation (those cost 15$ and you have a lot of power)

put the DC/DC converter close to the camera, the power loss in the cable is not negligeable (camera won't power up or will shut down during operation)

Face and Body Scanning / Re: More megapixels or better optics?
« on: July 14, 2014, 04:08:39 PM »
as marcel suggests the L has superior optics, i borrowed the lens so do not have it to do another comparison, the one in question is very expensive so a middle of the range 50mm fixed focus may be better, once i get the lens again i will do a fair comparison between that and the kit lens, if i remember when i used it before i was just starting and used different lighting between the two and raws for the better image, however just looking at the images its obvious the L lens produces better images

Comparaison would be great, thank you very much :D
Could you post screen shot of the resulting 3D model and a full resolution pictures of the same thing with both lens?

Face and Body Scanning / Mask unfocused area?
« on: July 01, 2014, 03:35:19 PM »

I I have XXX pictures of a face, and sometimes some part are out of focus,
example, side pictures right eye in focus, left eye unfocused.

Would I get better result by masking the lef eye?


Face and Body Scanning / Re: More megapixels or better optics?
« on: June 30, 2014, 11:38:59 AM »
I started with a 600D canon which is an entry level dslr, managed to get an 'L' lens and the difference in noise and quality of scan between this lense and stock lense is huge. Decent optics are a must, i will do a comparison eventually with a 5DMk2 and 600D using the 'L' lens to see what really matters. I will upload some test images showing the difference between the cheap 18-55mm kit lense and the L lense, both shot in my living room with no lighting

Could you give more information about this L lens? I never touch a DSLR before and
plan to buy a 600D.

Face and Body Scanning / Re: More megapixels or better optics?
« on: June 23, 2014, 12:41:59 PM »
I can't awnser your question however I'd loke to warn you about something else,
do not make the same mistake I did.

Low end point and shoot camera do not allow ISO/aperture/tv-speed (shutter-peed) management.
Most of them choose for you. so if you have 20 cameras, it is most likely that they will all choose something different.

The canon a2500 is perfect because you can use the CHDK firmware which will allow tv speed, and ISO to be choosen by the user (I don't know about aperture).


I have 21 Coolpix L27 and 9 times out of 10 I have 2 of them (never the same) which
are out of focus.
2 of them are focusing on the the cameras which are behing the subject.

This can be solved by doing 10 scan and 1 of them should be OK.
The user guide say that the camera focus on the center of the image, here it's clearly focused on a corner.

However I do not own the pro version so I guess it's out of my reach.

After several other experiment it seems that subject 2 has a larger head than 1, so moving backward the camera (2cm) solve the problem.

you think lens distortion is the problem?
the source files are here :

Pages: [1] 2 3