A Few Questions Graphics Card and Camera Lens related

[digital]

Will a GTX 760 4GB 256-Bit GDDR5 PCI Express 3.0 Graphics Card work?

My system
Core i7-4770K
32Gbs 2400Mhz G-Skill Ram

I'm buying a Nikon D3200. It comes with a 18-55mm f/3.5-5.6 AF-S DX VR NIKKOR Zoom Lens.

I want to take pictures of static objects like boots and shoes and head busts inside a white light tent. Will the Nikon's stock lens be good enough? If not what type of lens should I get?

[Mr_Curious]

No... well, not at least in my experience.  The stock lens that comes with the Nikon D3200's are of pretty poor quality.  I was unable to get truly pin sharp images with them.  I had to replace all of the stock lenses with better quality, prime lenses which made a huge difference in overall image quality.

Cheers

[digital]

Thanks for the reply Mr_Curious

After reading a few posts on this forum related to the D3200 and some other posts from Lee of IR. I'm thinking the extra cost for a Canon T3i/ 600D might be the better option.

I'm still not sure of what lens I should seek out. I think Lee uses a 50mm, but the subjects he takes pictures of are further away from the camera than I plan to be.

Is there enough power in the graphics card I asked about to use the ultra settings?

I use zbrush all the time, so this 3d photogrammetry stuff is very exciting stuff to me! I just want to make sure I set myself up well on a small budget for my first jump into it.

I'm mostly confused about the camera stuff right now.

[Marcel]

The closest focus distance for the Canon 50mm f1.8 is 0.45m (1,5 feet), so you can get pretty close to your subjects. If you would like to do really small objects (like jewelry) then you might need a different lens.

Personally I would go for the Nikon D3200, because it has a significantly higher resolution (24MP instead of 18MP). Provided you can make perfect pictures (using a tripod and good technique), Photoscan will like this extra resolution.

Your GPU doesn't matter that much. Processing times at Ultra setting are measured in hours anyway. You could get a faster GPU and it might be 20% faster, but it just means you are waiting one hour less. Generally you will run the Dense Cloud overnight anyway, so as long as it is done in the morning it's fine

If your motherboard has 8 DIMM slots that would be more flexible (so you can extend to 64GB in the future if you need it)

[Mr_Curious]

Personally I would go for the Nikon D3200, because it has a significantly higher resolution (24MP instead of 18MP). Provided you can make perfect pictures (using a tripod and good technique), Photoscan will like this extra resolution.

Respectfully, I would completely disagree with this statement. You know, that's what I thought too, until I actually purchased and tested the D3200 against other "similar" offerings by Canon.  Don't believe the megapixel hype... because that is what it is.... hype.  The Nikon D3200 in both my considerable testing and experience are actually vastly inferior to the lesser megapixel Canon's.  Don't believe me, test them side by side for yourself and you will see... both Lee (INFINITE) and I have done just this and the results speak for themselves.  I know, this seems counter intuitive but the truth is that until you use these things in production and test under real world conditions specifications put forth by manufacturers and marketing departments mean very little.  The proof is in the pudding.

At the end of the day, I'm not trying to change your mind about anything.  If you like using the Nikon D3200's more power to you.  I have a hybrid rig myself which consists of Nikons and Canon's so I have no axe to grind and I'm not being a fanboy.  I only mention this in the hope that I might save you some trouble and perhaps money.

Something else that you should also consider is the cost differential between Nikon and Canon external power supplies. The cost of Nikon power supplies for the D3200 is not trivial and if you have a lot of them, it really adds up fast.  While you can find good deals on Canon power supplies the same isn't true for the Nikon's.

Good luck and best wishes in whatever decisions you make.

Cheers,

MC

[Andrew]

I too had pretty bad experience with Nikon D3200 - it produced VERY soft images.
One thing to take into consideration - D3200 is actually last year's model, now replaced by D3300 with low pass filter removed, much like D7100 that I use and am very satisfied with.

However, I can't say that 24mpixel worth of OLP-less imagery is noticeably better than 600D's. They are really, really close, which puts Canon at clear advantage - less pixels means less computations and less memory to produce virtually identical scans.

-Andrew

[Marcel]

Could it be that the difference in sharpness between Nikon and Canon is due to the in camera JPEG conversion? The thread starter 'digital' is going for a single camera setup, so he can (and should in my opinion) photograph in RAW.

I have not compared these cameras myself, but I would be very surprised if actual sharpness is substantially lower on the D3200  under perfect conditions. I just checked some sample shots of the D3200, and they look tack sharp at 100% (unfortunately it's on a subscription-only website, otherwise I would show the image). I can also imagine it is an autofocus issue (in which case the D3200 would be a bad choice because you need reliable autofocus for a single camera setup).

Whether the extra resolution is useful probably depends on the subject matter as well. I can imagine that for face and human scans you don't gain that much detail (especially if you process at 'High'). But from our experience with scanning static objects we see a huge amount of extra detail between scans photographed with a 5D MKIII and a D800. But we are lucky to have subject matter that has a lot of micro detail which makes it very suitable for scanning, so what is true for us might not be true for you. Photogrammetry is tricky business

[Magnus]

Hello!

Maybe this can be a little bit of help.

On dpreview you can compare a bunch of different cameras and also download the raw files.

Here is a link to "Studio scene comparison (RAW)" for Canon 600D (a few pages back you can see the jpg comparision).

http://www.dpreview.com/reviews/canoneos600d/18

I personally have some Canon 1100D, 600D and 100D and also some Nikon D3200. My favorite is the 100D (1100D is also great for the price).

Best, Magnus.

[Mr_Curious]

Interested in the D3200?

Then I would suggest that you read this thread very carefully.... if you've not already.

Here:

http://www.agisoft.ru/forum/index.php?topic=1411.0

This topic has been discussed in great depth previously.

Cheers,

MC

[David Cockey]

I use a Canon 500D/T1i and one of the lenses I use is the Canon 18mm-55mm zoom lens which came with the camera. While the lens may not be quite as sharp as some of the other lenses I use PhotoScan has gotten excellent results with photos taken with that lens.

Another lens I use is the Canon 40mm f2.8. While this lens is compact and relatively inexpensive it is very, very sharp; particularly at f5.6 to f11 which is where I usually use it. The 40mm focal length (64mm 35mm equivalent) has worked well with PhotoScan.

[Marcel]

Interested in the D3200?

Then I would suggest that you read this thread very carefully.... if you've not already.

Here:

http://www.agisoft.ru/forum/index.php?topic=1411.0

This topic has been discussed in great depth previously.

The problem with this discussion is that people test the D3200 with a kit lens, and then conclude that it is a bad camera because the resulting images are soft.  The conclusion should be that the combination of the camera and kit lens give bad results.

The comparison between D3200 and Canon 600D (http://www.agisoft.ru/forum/index.php?topic=1411.msg7296#msg7296) is completely unfair because the D3200 is using a crappy 18-55mm kit lens, and the 600D is using a 50mm F1.8 (one of the cheapest but sharpest lenses around).
This is comparing lenses, not cameras.

Here is a crop of a shot with a D3200 with an extremely good lens:


Making perfectly sharp photos with 21+Megapixel camera is just really really hard. I saw some other users do some indoor handheld test shots with a kitlens and then say that the camera sucks. To do a valid comparison of the cameras you need to take RAW photos of a well lit static scene with a high quality 50mm prime lens and overlay the photos so you can look at the pixel level. Any other comparison will lead to wrong conclusions.



Also, Lee showed a really bad D3200 shot in a comparison:

https://www.dropbox.com/s/w2lfk5re1sopcgx/D3200vs600DvsD7100.jpg

If you look closely, you can see that this is not lens softness but camera movement (the blur goes from left to right, lens softness would be an overall softness). I am pretty sure this is caused by mirror slap from the camera, not lens softness or a bad camera.

Mirror slap can cause problems like this, but it is a very tricky problem to diagnose because it depends on shutter speed, camera and lens weight, camera orientation, etc. The best way to prevent it would be to use Mirror Lockup (but I am not sure if that is doable in a 100+ camera setup).
So it might very well be that in the lightning situation Lee uses the Canons work fine, but the D3200 create blurred images just because they happen to use a shutter time where the mirror slap causes blur.

So my overall point is: don't compare results between cameras unless you use exactly the same lenses and a perfect technique.

[Mr_Curious]

Right Marcel,

So you've made my point exactly, which is the D3200 sucks for photogrammetry.

You are 100% correct it is the poor kit lens is mainly at the root of the problem with the soft focus.  I've tested this camera extensively, both with the kit lens and with extremely good prime lenses and YES there is a huge difference in image quality... the kit lens is rubbish.

Is it possible to make a good photo with the D3200 -- yes, absolutely if you have the right lens and laboratory like conditions.

Again, as you point out the D3200 seems very sensitive to mirror movement and for whatever reason it still makes less sharp images than the canon under typical conditions that would be used for photogrammetry..... so my conclusion is this....

With the Nikon D3200 I can't use the kit lens... so I have to go to the added expense of purchasing a new one.

That the camera may or may not take pin sharp images.... due to mirror movement.

So... it sounds to me like the D3200 still isn't a great choice for purposes of photogrammetry... which is the whole point of this discussion.

Again, I'm not trying to force anyone to believe my opinion.  Get yourself a 20+ D3200's and use them in your rig... then draw your own conclusions.

Cheers,

MC



 

[Marcel]

If you are talking about an epic multi camera rig then the D3200 might not be suitable (because of less good in-camera JPG conversion, bad kitlens and expensive chargers, etc). But the topic starter is gong for a one camera setup. I think he'll benefit from the extra resolution (without needing lab conditions for teh photography).

So I think we are arguing slightly different points here, but we basically agree with eachother.

I definitely feel your pain about purchasing a bunch of D3200s. I purchased 2 Sony A7R cameras and I had a similar bad experience (massive shutter shock ruining scans). Even though I could return them to the shop with minor losses, we wasted more than 3 weeks redoing stuff.
I really admire you guys with the huge setups.

[Mr_Curious]

So I think we are arguing slightly different points here, but we basically agree with each other.

Yes, I fully agree we are on the same page.

And yes, at least in my case I am talking about use in a multi-camera setup.

One thing is for sure, the kit lens that comes with the D3200 has to go :-)

Cheers,

MC

[digital]

Thanks for all the information and personal insight guys. I read the "Most reasonable DSLR for multi-cam face capture setup" a few hours after my original post in this thread and I'm glad Mr_Curious directed others there. That thread is very insightful.

I would have posted here sooner, but I thought it might be better to see what everyone had to say before chiming in.

Earlier Nikon models let you lock the mirror by going into Live View. YT says The D3200 disabled this feature.

I'm looking at this as a money issue. What will save me the most amount of money while giving me best results possible for the price. For right now the answer seems to favor Canon.