2
\$\begingroup\$

I took a panorama picture with my Google Pixel 6 (default camera application + Android 12):

enter image description here

If one looks a bit close to it, the image is a low-quality. E.g., zooming-in on the top left corner:

enter image description here

How I can improve the quality of this panorama picture?

Improve this question
\$\endgroup\$
3
  • 1
    \$\begingroup\$ The scene looks quite dark so it is very likely that your camera (which I guess was on auto mode) automatically raised the ISO setting quite high. Mabye the JPEG compression also got involved in that noisy output. How do you expect to improve the quality? You can try to reduce the noise with software but you will loose detail in the image. \$\endgroup\$
    – MrUpsidown
    Commented Jan 12, 2022 at 15:19
  • 1
    \$\begingroup\$ IMO, if you want good quality you should 1) Use manual settings on your camera (low ISO, longer exposure time) 2) If possible shoot in raw mode (no "in camera" JPEG compression) 3) Use a tripod for that kind of dark scenes where you will need longer exposure times. You should get better quality images in the first place! \$\endgroup\$
    – MrUpsidown
    Commented Jan 12, 2022 at 15:22
  • 1
    \$\begingroup\$ @MrUpsidown Thanks very much for the great advice, all points noted! \$\endgroup\$
    – Franck Dernoncourt
    Commented Jan 13, 2022 at 3:15

2 Answers 2

Reset to default
3
\$\begingroup\$

First there is no image w/o noise, even in perfect light you can see kind of noise.

For me this is combination of noise, compression and algorithm of Google to stitch photos and postprocess them. I see two possible ways to improve the photo.

  1. In postoprocess (from existing image) with Lightroom, Darktable, etc. software you can try to reduce the noise and smooth the texture in sky. You can play with different parameters for noise reduction and details.
  2. Because Android camera application do not give you much controls you can try to change the entire workflow. Start with taking photos many in RAW (so maned PRO mode in Android) mode. Do not forget the images must overlap around 25% to be able later easy to stitch them. Then edit images (same settings to apply to all images in sequence) and stitch them. The process become quite long but you can get better results.

P.S. To get less noise you may need to set ISO in PRO mode to the base ISO and set the phone to tripod to get better alignment and deal with lower shutter speed

Improve this answer
\$\endgroup\$
1
  • 1
    \$\begingroup\$ Thanks for the great answer! I'll look into both points. \$\endgroup\$
    – Franck Dernoncourt
    Commented Jan 12, 2022 at 7:19
3
\$\begingroup\$

Romeo's answer covers all the bases, but I thought I'd throw in a practical example, using a DeNoise plugin - this one from OnOne.

Of course you will get better results from a RAW image or indeed just a larger jpg, so this isn't showing it off at its best.

enter image description here

I set this to the minimum necessary to kill the noise, then tried sharpening a little to try grab some detail back. You can immediately see that; yes, it's successfully killed the noise, however it has completely lost any detail from the smoother areas of water near the centre & it's not really doing the sky any favours either.

I wouldn't actually let that put you off trying out the product. It is remarkably good when it's got a bigger/better image to start with.

The only real competition I know of in this area is from Topaz. Both are worth a look at. I tried a Topaz demo a while ago & it was good, but I couldn't justify having both.
There's a comparison guide at Petapixel - ON1 NoNoise AI Versus Topaz DeNoise AI: Noise Reduction Throwdown

No doubt Photoshop will come up level eventually [& perhaps even has already] but my Mac is too old to run the latest AI-enhanced versions, so I cannot test.

Improve this answer
\$\endgroup\$
1
  • \$\begingroup\$ The other known noise reducer is NeatImage and has one version that is free for personal use. ni.neatvideo.com I think it does not use an AI, but it has been there for a while and produces good results. \$\endgroup\$
    – Rafael
    Commented Mar 2, 2022 at 1:48

Not the answer you're looking for? Browse other questions tagged or ask your own question.