Which DSLR camera to choose for gigapixel photography

In search for which camera to choose for gigapixels: Crop with less gigapixels? or Full Frame with more gigapixels? I made a simple comparison table where pixels per degree are calculated:

crop factor angle in degrees
300mm 1.5x (APC Nikon) 1.5 3.1
300mm 1.6x (APC Canon) 1.6 2.9
300mm 1x     (FF) 1 4.6

Assume, we have 300mm lens and portrait orientation of camera:

Camera Megapixels image width in pixels crop angle in degrees pixels in 1 degree Number of images required to cover 360 (with 0 overlap) Equatorial 360 panorama’s width in pixels
Nikon D7200 24 4000 1.5 3.1 1290 116 464516
Canon 750d 24 4000 1.6 2.9 1379 124 496552
Canon 7d mkII 24 3648 1.6 2.9 1258 124 452855
Canon 5d mkIII 22 3840 1 4.6 835 78 300522
Nikon d810 36 4912 1 4.6 1068 78 384417
Canon 5ds r 50 5792 1 4.6 1259 78 453287

These number doesn’t take into account diffraction which can be a practical issue.

Sony APS-C cameras with 24Mpix should be treated similar to Nikon D7200.
More pixels in degree – better. FF is better than APC-S (requires less images). So for me the winner is Canon 5ds r – it allows to achieve almost the same pixel density as other cameras but you can get the same gigapixel with less images – which will mean  less overlaps, less space on a card and computer required, faster shooting/stitching..

Samsung Galaxy Note 4 vs DSLR

Just made an experiment how my new Galaxy Note 4 stands against my DSLR Canon 5d mark II.
I zoomed 5d mark II with 16-35 f/2.8 II L equal to Note 4  horizontal FOV (what resulted in 29mm) then made several shots from the same point. 5dmk2 JPG+RAW (as I always shot). JPG: WB – auto, Picture style – standard, quality – normal, Aperture priority f/8, evaluate metering.
RAW: some standard tuning to my taste, shadows /highlights – additional sharpness CA removal.

Note 4 camera: h16lsha00sm (according to *#34971539#)
Here is the final result picture (all the images zoomed to actual pixels):

Have to admit what Note 4 is very close or even better to 5dmk2 in Jpeg (what is not surprising as DSLR normally don’t add too much contrast/saturation/sharpness to JPEG engine), And 5d mk2 in RAW is all way better than Note 4 (thanks god :))

Dinner at Sugar Factory, Las Vegas

professionalpanorama.com meet airpano.com and aerialphotoimage.com during dinner during IVRPA  2014:)

Las Vegas at dusk spherical panorama from Eiffel Tower

Las vegas dusk 360 panorama

Resolution: 14000×7000

Camera/lens: Canon 5d mark II und 8-15 f/4L at 15mm.

Software: PTGUI 10.0.10 Pro,  Photoshop CC 2014

Computer: Intel I7 4.1Ghz und 24GB of RAM

As people request some info about panorama: It was shoot with Canon 5d mark II und 8-15 f/4L at 15mm. Stitched from 103 photos (33*3 bracketing + 4 separate photos of fountain . ISO 100, f/11 (2.5s, 0.6s, 1/6s) bracketing (without tripod:).

Initially I ‘ve made easy version – tonemapped all 33 sets with SNS – then stitched pano from that TIFFs. It took 2 evenings But wasn’t pleased with result – looks too “poisoned” to me. BTW: here is it: http://professionalpanorama.com/2014.05.29_las_vegas_paris_eiffel_panorama_hdr/

When I tried process which is my favored: PTGUI exposure fusion with layers and started to work with generated PSB file. The PSB has exposure blended layer and all the layers with black masks on top of it. As it is shoot from different points – many parallax errors were present – to fix them I normally finding layer with required part of image and required exposure (as it is bracketing) and reveal required part of it by painting white on a mask. If you shoot from tripod with panohead – then all layers are alligned – and revealed part is exactly aligned to the blended panorama, but in case of the building, you shot from different points  – and parallax errors occurs. To fight with them i normally copy part of required  layer on top of everything – and transform/allign/warp, As it was shoot during 18min – lights changes at dusk time dramatically – I had to apply adjustments on each copied layer (levels, color balance). Some layers have 5-6 clipped adjustment layers. At the end the result was what whole image was copypasted with adjustments so I deleted ptgui exposure blended part – the image was without holes – which mean I had manually created whole pano :)

The whole process took about 3 weeks (several evenings/nights per week)

Screenshoot of photoshop (somewhere in half “job done”)


Me (with a black backpack) during the shooting (thanks for photo to  Sergey Semenov (airpano.com))