Pages: [1]   Go Down

Author Topic: Shift, Stich or keystone correction - four samples  (Read 1193 times)

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11310
    • Echophoto
Shift, Stich or keystone correction - four samples
« on: March 20, 2016, 04:09:31 am »

Hi,

Here is an experiment I conducted a few days ago. Shot a church from same position with:

- Canon 24/3.5 TSE LII in vertical with shift and merged three exposures in LR (this was quick and dirty)
- Used same lens in horizontal position and and merged two images using vertical shift on the TSE lens
- Used Canon 16-35/4 zoom at 18-20 mm (I guess) on the HCam Master TSI adapter with some vertical shift
- Used the Canon 16-35/4 zoom at 16 mm and did keystone correction in Lightroom. That gave me a cut off right lower corner that I fixed with content aware fill in LR.

http://echophoto.dnsalias.net/ekr/Articles/Shoots/FourWays/

Raw files included. I don't know how "raw" those files are, really, as they are stitched/mangled in LR CC, but they should contain a reasonably close to raw representation.

Best regards
Erik

« Last Edit: March 20, 2016, 04:14:12 am by ErikKaffehr »
Logged
Erik Kaffehr
 

marc aurel

  • Full Member
  • ***
  • Offline Offline
  • Posts: 150
Re: Shift, Stich or keystone correction - four samples
« Reply #1 on: March 20, 2016, 06:41:38 am »

Hi Erik,

thanks for that interesting comparison. A few things that I noticed:
  • You cold get away with all these images for most purposes.
  • As expected for all methods the top of the tower is the weakest spot.
  • Resolution and detail of the TS-E with one shot level and one shifted vertical (File 4473) has a lot of detail and resolution. To me it looks as if the tower could possibly be a bit sharper if focused a bit more distant (field curvature).
  • The keystone corrected example (File 4462) is taken at f8, the others at f11. That could disadvantage it a bit. But even at f11 I doubt that it could match the clarity of detail in the foreground (grass) of the other examples, especially ot File 4473. By tilting the camera you tilt the plane of focus which puts foreground further away from the plane of focus. To me this image has the weekest image quality of the four.
  • The stitched 16-35 with the HCam Master TS looks very good in most part of the tower. I see a sudden drop of sharpness when it comes to the rooster on top (File 4465).
  • The lack of detail in the far corners makes it difficult to guess how image quality would be in the outer regions of the image circle.

Very interesting stuff.
Regards - Marc
« Last Edit: March 20, 2016, 06:46:27 am by marc aurel »
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11310
    • Echophoto
Re: Shift, Stich or keystone correction - four samples
« Reply #2 on: March 20, 2016, 04:29:34 pm »

Hi Marc,

Thanks for chiming in! And thanks for digging into the pictures.

Just two small remarks:

1) The image shot with the Master Cam HS is only shifted, not stitched

2) I feel that the Canon 16-35/4 has optimum aperture around f/8. With my sample of the 24/3.5 TSE LII I feel that I need to stop down a bit more for best sharpness, especially when shifted. Both lenses are essentially brand new. I was considering to send in the 24/3.5 TS for repair, but I would guess that it is "within tolerances" so I didn't care.

I have noticed loss in sharpness and colour fringing on the rooster on top. I have taken some liberties with the image circle on the 16-35/4.

What I may have found is that I was most happy with the vertical stitch on the TSE and with the shifted 16-35/4.

Just to explain, the idea with this posting is mostly to give some "food for thought" on software corrections vs. shift and also stitching.

Best regards
Erik

Hi Erik,

thanks for that interesting comparison. A few things that I noticed:
  • You cold get away with all these images for most purposes.
  • As expected for all methods the top of the tower is the weakest spot.
  • Resolution and detail of the TS-E with one shot level and one shifted vertical (File 4473) has a lot of detail and resolution. To me it looks as if the tower could possibly be a bit sharper if focused a bit more distant (field curvature).
  • The keystone corrected example (File 4462) is taken at f8, the others at f11. That could disadvantage it a bit. But even at f11 I doubt that it could match the clarity of detail in the foreground (grass) of the other examples, especially ot File 4473. By tilting the camera you tilt the plane of focus which puts foreground further away from the plane of focus. To me this image has the weekest image quality of the four.
  • The stitched 16-35 with the HCam Master TS looks very good in most part of the tower. I see a sudden drop of sharpness when it comes to the rooster on top (File 4465).
  • The lack of detail in the far corners makes it difficult to guess how image quality would be in the outer regions of the image circle.

Very interesting stuff.
Regards - Marc
Logged
Erik Kaffehr
 
Pages: [1]   Go Up