Forum

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 - Coadey

Pages: [1]
1
Face and Body Scanning / Re: Samsung NX1000
« on: April 04, 2013, 12:35:20 AM »
With the Arduino it would be possible to switch between remote trigger and USB host functions (via a shield) to download images, but I haven't explored that yet.

This is what I was kinda thinking as well..(guess you are basically talking about switching the resistor in/out ) although I'm not too sure if the camera is going to like this very much...the other little problem is that the camera needs a button press to start pc connection unless there is a way round this ?
Right, while the camera tolerates the USB switch, it does need user interaction to choose "computer" versus "printer" on the host connection.  Upon switching back, it defaults to playback, so a shutter press is needed to get it back to capture mode.  An actuator could do the button press but I don't really want to go that far yet.

A true remote capture with a Nikon is going to be better for a serious workflow, but unfortunately at 2-3x the cost. I'm hoping Samsung steps up their remote app or we get a CHDK-like option.

2
General / Re: [Help]Any tips on scanning a model in her wedding dress
« on: March 31, 2013, 08:43:16 PM »
Hi koyko4, Did you try the scan?  Any conclusions or tips?

It's a common issue scanning objects that don't have enough complexity, present specular highlights (like silk or satin) or incorporate repeating identical patterns.

I've had some success using adhesive targets (not the best idea on a wedding dress).  If you can get really sharp, high-res/close-up photos, the fabric itself might provide enough texture.  Lighting it right seems to be key, especially lots of diffusion.  Do you have a studio?  On clothing that moves it's a real challenge with a lot of post cleanup needed from what I've seen.

3
One is the inherent sharpness of the lens for which there is an optimum f-stop. A higher numerical f-stop will reduce this part of sharpness.

I just went through this exercise on my NX1000 kit lens, and David makes a good point; each lens has an optimal f-stop for a given focal length to achieve maximum sharpness (often about 2 stops down):

http://en.wikipedia.org/wiki/Stopping_down

I believe this is your lens:

http://www.photozone.de/canon-eos/404-canon_1855_3556is_50d?start=1

At high stops you'll have everything in focus, but with lower sharpness, more aberrations and probably more noise due to higher ISO.



4
Feature Requests / Re: Cut and trim by masks
« on: March 31, 2013, 07:59:01 PM »
+1 for incorporating masks into geometry.

It would be really helpful if Photoscan used image masks to assist in model creation (or at least clipped vertices outside the masks after creation). 

5
Face and Body Scanning / Re: Samsung NX1000
« on: March 31, 2013, 07:20:57 PM »
I'd sort of forgot about the NX1000, instead investigating a mixed chdk point&shoot / DSLR solution. I'll have to look at this again.
I settled on the 20 MP NX1000 versus other APS-C mirrorless models as it seems to be a sweet spot in price vs IQ/performance for ?260.  The Sony NEX-5R with its 16 MP sensor was a close second, with its better ISO, but at ?460 it was less ideal (and can only use an infrared remote).  The Nikon D5100 uses a similar 16 MP Sony sensor and supports tethered USB, so it could be a good option too at around ?400.

Getting the images back to a central location remains a big problem for the multi-cam setup...
With the Arduino it would be possible to switch between remote trigger and USB host functions (via a shield) to download images, but I haven't explored that yet.  I only have the Photoscan Standard license which can't script model creation, so there's still manual work involved anyway.

I'd love to see some of your scans, especially how you deal with subject movement.
I've tried to dampen as much inertia as I can with high ratio gear boxes, but I haven't been able to eliminate it fully or turn the table slow enough to support continuous capture (which seems to be ideal).  I need to rebuild the table now that I've hashed out the limitations.  Of course the entire scan has to be done as quickly as possible so the subject doesn't fidget, and sitting is better than standing.  How is everyone combating this?  I'll try to post some scans at some point.

Thanks for the feedback. :)

6
Face and Body Scanning / Re: Samsung NX1000
« on: March 30, 2013, 09:46:02 PM »
I just finished building a rig from scratch for around 600 euros. I decided on a Samsung NX1000 and built a motorized turntable both controlled from an Arduino.  The platform has 28 marked positions picked up by a photocell that triggers a capture and rotates between positions automatically, with a chroma key screen for scripted masking.

I've learned a few lessons about the NX1000 in this setup so far:
  • With the 20-50mm kit lens, the highest MTF and least flare is at f5.6 or f8.
  • In Single Drive mode, full manual and SuperFine JPG, photos take 750ms to capture (including 200ms shutter press) on a UHS SDHC card (SanDisk Extreme Pro 95 Mbps).
  • In Continuous Drive (low) mode, it can grab 12 photos in 3.5 seconds, then pauses 7.2 seconds to write them out.
  • I need to use 4 x 400W halogen lamps with soft boxes to get decent exposures with continuous lighting.
I'm getting pretty good scans with just a single camera (around 40,000 points on a person), but need to make some improvements. It seems 2 cameras would be optimal to capture finer details. Here's what the trigger circuit looks like (pretty easy to adapt to trigger multiple cameras at once):



The Arduino sketch just needs this in the setup():

Code: [Select]
int cameraTriggerPin = 45;
pinMode(cameraTriggerPin, OUTPUT);
digitalWrite(cameraTriggerPin, LOW);

And you trigger a single capture with:

Code: [Select]
digitalWrite(cameraTriggerPin, HIGH); // Trigger camera
delay(200); // Wait 100ms+ for camera to detect trigger
digitalWrite(cameraTriggerPin, LOW); // Stop trigger
delay(550); // Wait for camera processing to recover

You could wire up the D- pin if you wanted to trigger auto-focus separately, but the NX1000 will focus before it snaps a photo anyway (if AF is enabled and AF Priority is on - but I recommend manual focus). 

Practically all Micro USB cables and extensions (USB "On The Go" / OTG) out there only have 4 wires connected through.  I've found this is the best place for 5-wire cables:  http://www.usbfirewire.com/uextendmicrob.html

Here are the other parts to do this solderless:

http://proto-pic.co.uk/usb-microb-plug-breakout-board/
http://proto-pic.co.uk/n-channel-mosfet-60v-30a/
http://proto-pic.co.uk/68k-ohm-1-4-watt-resistor-pack-of-20/
http://proto-pic.co.uk/breadboard-mini-self-adhesive/
http://proto-pic.co.uk/solderless-headers-10-pin-straight/

As Mala pointed out, the USB remote trigger works in modes other than WiFi, so you're limited to capturing to the SD card.  But using a trigger is much better than the app-controlled WiFi Remote, since that mode has practically no control (full auto).

I really appreciate seeing all the work others have done.  I can share more about my setup if anyone's interested. :)

Edit:

Here is one way you can adapt this to trigger multiple NX1000 cameras at once.  Please note I haven't tested this, and if you're using long USB cables the voltage drop might prevent it from working.  Feedback appreciated:



Coadey

Pages: [1]