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 - Andrew

Pages: 1 ... 3 4 [5] 6
61
I promised to share sample photos from my test multi-cam capture for IQ comparison between Canon 600D, Nikon D3200 and Nikon D7100, but I can't get my hands on the files right now (damn you, holiday season!), so I just took a few snaps of my coleague here.

All three cams were set to full manual mode, f11, hand held at 1/160s and no image-altering shenanigans, with kit zoom lenses at around 50mm. Used camera's built in flash units, asked the guy not to move around too much :) He doesn't like internet fame so I cropped the images. All in all, not very scientific test but still should give an idea of how these cams compare.


Nikon D3200 with AF-S Nikkor 18-55mm: https://www.dropbox.com/s/raz9e48o7rjkw0v/D3200.JPG

Canon 600d with EF-S 18-55mm: https://www.dropbox.com/s/mjlyc4ejj38lvw5/600D.JPG

Nikon D7100 with AF-S Nikkor 18-105mm: https://www.dropbox.com/s/g7zr8tlh4slwimg/D7100.JPG

62
Going away on a three-day weekend now, but after I return I can share sample photos from multicam setup (600D, D3200 and D7100), if you'd like to compare against your results. I honestly think that D7100 is as sharp as 600D (even has a tiny edge due to more pixels), definitely in a different league than D3200, so perhaps your unit had faulty focus or my kit 18-105lens was sharper, who knows.

-Andrew

I tested a D7100 and there was very little different between it and the D3200. I'm not sure where the razor sharp comes from?




63
Lots of valuable insight gentlemen, thank you!

One thing that surprised me a lot Lee was that you put 600Ds above D7100 and D800/D800E in terms of resolved detail, wow!

D3200s are insanely soft, true, but my D7100 looks razor sharp (right up there with 600d) and although I haven't had the pleasure to work with D800/D800E, the resolution charts from these babies look insane!. Perhaps at higher f-stop light diffraction inside the lens (and we rarely use high end glass on our multi-cam setups, do we) decreases resolved resolution to levels where the 36mpix sensor doesn't matter anymore.

One additional thing to note is that I have witnessed numerous and serious build quality issues on higher end Nikons. My D7100 saves photos to cards erratically (every now and then the file is corrupt), that frigging custom usb port has gotten loose after less than 10 times I plugged the cable, and yeah, cable length is a joke. From what I hear, there are lots of units with faulty focus (tendency to left-focus and such), and don't get me started on oil spills soiling the sensor...

Long live 600d! ;)

Cheers,
Andrew

64

Nikon D3200 seems to be the camera of choice these days, replacing Canon 600D, with main rationale behind it being price advantage over 600D, but also, to quote Lee on FB, "These things are as quick as gunslingers! Very fast and perfect sync compared to 600D's".

Now, I understand the price factor all too well, but speed? Can anyone (Lee? :)) explain what that means specifically with respect to 3D scanning?

With devices like Shutter Cell/Camera Axe/Multi Flash Plus, one can precisely synchronize cameras with flash units, no matter how big the shutter lag is (btw. D3200s shutter lag is larger than D600s), as long as shutter lag is consistent. To me both D3200 and 600D seem to have consistent shutter lags but I didn't have time/means to do very precise tests yet - I can only confirm they are both light years ahead of old Canon 550D.

So:

1) Is D3200's shutter lag much more consistent than 600Ds?

2) Are there any other important speed/synch factors I am missing?

3) Isn't D3200 a tad soft for precise 3D reconstruction? Does sharpening photos help?


Oh, just to be clear, this is all for internal indie studio use, I am not about to offer scanning services and poach anyone's business :) Also, I only do occasional scans so convenience (battery bay doors blocked by camera mount etc.) are of secondary importance, although it would be nice to hear all known pros and cons.

Thanks in advance!
Andrew

65
Face and Body Scanning / Re: Synchronizing different DSLR cameras
« on: July 08, 2013, 06:11:43 PM »
I just dug up useful article testing flash duration for various speedlights as well as two studio flashes.

http://www.gock.net/2012/01/flash-durations-small-strobes/

Yongnuo YN560 at its weakest setting emits light in 43 microseconds, while "Mystery" Ebay light (probably close to the cheap Chinese gear I got), takes 7120 microseconds to flash :) That's a cool 165 times slower, wow! No wonder I wasn't able to measure lags with too much precision :)

66
Face and Body Scanning / Re: Synchronizing different DSLR cameras
« on: July 07, 2013, 07:49:30 PM »
Hey there, glad you shared my post from the other forum, it was way past 2am when I concluded my testing, felt too tired to post here as well :)

As for the details of my testing - yes, I did test shutter lag for individual cams by placing them directly in front of flash unit and adjusting flash delay until I got uniform illumination of entire frame, just like Maurice did in his blog post. The strange thing is - whenever he saw his camera's shutter partially open, he would get a photo where there was clearly defined border between lit part of the frame and black part obscured by shutter. In my testing though, I never got this crisp border - I would always get images slowly emerging from darkness, as I was increasing flash delay by each millisecond. Instead of border, I would get dark gradient lifting from the frame. Could it be because of my cheap studio flash? Perhaps it isn't as quick to fully light up as Maurice's small speedlight?

Now my findings: Nikon D7100 understandably wins by a large margin - shutter lag was fast and consistent, around 54ms if memory serves. The other Nikon, D3200 landed last, and I mean last (145 or 150ms I believe), but it too was rather consistent. Canon 600D's were in the middle (90ms I think), while 550D was a tad slower (94-95ms), and a whole lot less reliable. You can get some black images or dark gradients as the shutter is dragging its feet.

I tightened my timings only to a point where I was able to shoot 1/100s with these 4 cams and flash units, with only occasional problems caused by 550D.

Andrew

67
Face and Body Scanning / Synchronizing different DSLR cameras
« on: July 04, 2013, 05:23:16 PM »
I just bought Camera Axe with Multi Flash Plus to solve different shutter lags on different camera models I have laying around (old Canon 550D, two 600D's, Nikon D3200 and Nikon D7100).

I set all cameras to Manual mode, set all zoom lenses to Manual Focus, made sure all cameras use same settings (ISO 100, F5,6, same WB), disabled any image altering shenanigans (like Active-Dlighting and some such), pointed all cameras at digital stopwatch and triggered them all simultaneously with Camera Axe/Multi Flash. I used no stobe lights to keep things simpler.

What I expected was evidence of consistent (down to 1-3ms) shutter lag, letting me adjust trigger times on Multi-Flash Plus unit, to achieve near-perfect sync.

To my surprise, I only observed consistent shutter lag on Nikon D7100 and Canon 550D. The two 600Ds would either sync perfectly with 550D or be quicker by 43ms, while Nikon D3200 would either sync perfectly with 550D or be slower, curiously by the same 43ms value.

All camera settings look the same, but perhaps there is something I'm missing, something that causes D3200 and 600Ds to sometimes pause and think before taking photos? Any ideas?

Cheers,
Andrew

68
General / Re: processing time
« on: May 31, 2013, 12:16:03 AM »
I haven't used 123D Catch for ages, but from what I remember it allowed maximum of 70 photos at maximum 3Megapixels (higher resolutions would be resized down to 3mpix by software while submitting to cloud. In contrast, Photoscan lets user decide the processing speed vs accuracy.

As for differences other than speed, there are too many to mention (disclaimer: I haven't used 123D since last Fall, they could have caught up by now), here's a few: Agisoft doesn't claim copyright to your work, Photoscan it allows mix of landscape/portrait photos, it allows exporting model for touch up and good UV layouting in external 3D packages and lets you re-project texture on a clean and efficient model, it allows ultra high quality results (provided that input photos are good enough and provided that you have enough memory and horsepower in your rig).

Cheers,
Andrew

69
General / Re: High setting = bad result, Low setting= good result
« on: May 17, 2013, 09:00:16 PM »
I often encounter problematic spots in my projects where lower quality yields better results. Makes me wonder if Photoscan could be able to determine problem spots where higher precision sampling produces inferior results and resort to using lower precision for those spots, to produce better overall result? Since errors in higher accuracy modes produce large triangles and Photoscan can detect those, it seems to be within the realm of possibility, doesn't it?

Andrew

70
General / Re: more detail and less noise
« on: April 26, 2013, 06:02:42 PM »
>use "separate" (instead of "group") camera calibration settings

This sounds interesting, I never tried this - are you referring to 'Split Groups' in left pane of 'Camera Calibration' dialog window? I usually use same focal lenght but sometimes resort to using OIS and wonder if this could improve precision of my reconstructions?

Thanks,
Andrew

71
General / Re: Benchmarking a GPUs
« on: April 21, 2013, 12:40:57 AM »
Wishgranter, you mentioned that latest ATI drivers seem to have fixed the VBO issues. Do you know if it is the case for all Windows versions (Win7, Win8 as well as Windows 2012 Server)?

It was also mentioned that Photoscan performs better (OpenCL-wise) on Windows 2012 Server than on Win7. Is it still the case using these latest drivers? What about regular Windows 8, is it as fast as 2012 Server or as slow as Win7?

Cheers,
Andrew

72
General / Re: nikon d800 E
« on: April 19, 2013, 12:55:29 PM »
For a few weeks now I have been using Nikon D7100, which lacks AA filter just like the D800. So far I have not ran into any banding/moire issues, and the extra pixel sharpness nicely helps Photoscan do its thing.

With D800E and it's sharp 36mpix, keep in mind that unless you use top shelf lens and shoot in optimal conditions (fast shutter, lens sweet spot aperture), most likely you will not see any benefits of using D800E over D800 (or cheaper 24mpix cams for that matter).

Cheers,
Andrew

73
General / Re: problems merging
« on: April 04, 2013, 02:24:45 PM »
Allow me to ask a follow up question then:

Since merging chunks degrades mesh quality, what are the best methods/tools/workflows for merging separately built mesh parts?

For simplicity sake, let's asume I use one PS chunk that I build three times, each time adjusting region cage. I end up with three exported dense meshes with some overlap. What is the quickest way to get rid of the overlap and merge tose three meshes into one smooth mesh?

Cheers,
Andrew

74
I just tried f11 on a rather smooth face and results seem better than f16, even though some parts of image escaped slightly out of focus. Per-pixel sharpness on low contrast areas seems to be really important. Thanks to the lens data from provided link I can easily see where my lens goes soft :)

Many thanks!

75
Thanks, RalfH! I realize the EF-S 18-55 kit lens is not the best piece of optical glass in existence, but if I want to extend my setup to more cameras, better grade lens will quickly kill my budget :)

With this lens, when shooting head shots, I am at about 80cm from the subject, shooting at 55mm f16 1/200s ISO100. This gives 12cm of focus (5.6 cm in front of focus point and 6.5 cm behind focus point), so almost entire head in perfect focus, for the price of slightly softer image overall.

To reiterate the point in my question: does Photoscan work better with slightly softer photos with wide DOF, or is it fine with slight blur in out of focus areas, for the price of crisp pixel detail in focused parts, or maybe something in between?

Cheers,
Andrew

Pages: 1 ... 3 4 [5] 6