I just caught the flu so it might be a lack of liquids, but you my good man are figuring this out fast. Feels like in no time the crappy noise filters will be a thing of the past!
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
iPhone Tweaking CameraSetup.plist to disable the agressive iPhone6 > noise filter
- Thread starter VooD
- Start date
- Sort by reaction score
Well, it may still take weeks to figre out how the stills need to be hacked - I've only managed to get rid of the NR in panos. I still don't have any info on the stills as, so far, all my experiments turned out to be futile.
How does Cortex Cam do it? Rumors are that he uses 5-10 seconds of video and then uses a multi-frame superresolution method to get the better resolution, so that might not help this problem. It seems that video and pano go through different processing chains than stills and for some reason allow more access to the NR parameters.
Apple, why did you ruin the iPhone camera by using this heavy-handed NR algorithm??????? Why don't you give us the choice of turning the @#$@#% thing off????
Keep searching Menneisyys2, and I reiterate: if you solve the problem and create an app for it, I (and probably a multitude of others) will gladly pay for it.
Must be somewhat similar to the pixel shifting method of the Pentax K3-II and the Oly EM5-II. (Haven't tested it myself yet.) I'll do it some time.
WRT pano, it's just that I managed to find out what to change. I'm absolutely sure traditional stills can also be "hacked" - I just haven't found out yet how, but will sooner or later find out. The fact that I could gain access to almost-RAW (apart from the fixed WB, the 8-bit color channels and the pretty heavy smearing caused by the panorama stitching itself) pano input shows there is a way of doing it for stills too. That is, there's a way of accessing the sensor output after applying the WB / converting the 12/14-bit color channels to 8-bit but before applying NR. And it's NR that is our biggest enemy because of its being far too strong.
Pretty much understandable they've gone this way. After all, the move from 8 to 12 Mpixels using the same sensor size easily results in noise increase. While the 12 Mpixel sensor uses at least a year newer technology than that of the iPhone 6, the sensor tech improvement in that year still can't offset the effects of the smaller sensor. Apple, therefore, needed to increase NR. Of course, it's a bit heavy-handed and unoptimized - but that's a problem with all manufacturers, not only Apple. For example, the Samsung Note 4 also uses very strong noise reduction too. (I recommend this article of mine on this subject: https://forums.macrumors.com/threads/shooting-in-good-light--get-the-best-image-quality-out-of-your-note-4.1840924/ . Samsung should also have fine-tuned their NR and use far less NR on lower ISO's. Fortunately, they at least allow for RAW sensor output on their newer flagships and have always strived for giving users as good camera hardware as possible - this is why their flagships' sensors have always been significantly larger than those of iPhones.)
If it helps, that's far faster than anyone else has lol. The issues existed in the hardware since 6+'s release but you are the first to get this far along. I would pay for a tweak to fix this if you release one!
I've just finished shooting and analyzing my long-promised (here in Finland, there is very little Sun during the Winter) well-lit low-contrast foliage tests. My hack allows for a LOT better image quality in very good light. Apple made a HUGE mistake by applying too much NR.
Of the several comparative images I've shot, let me show you the ones shot in front of the Frederik Pacius statue, Kaisaniemi park, Helsinki, Finland:
ISO64, iPhone 6s+ hacked (both gammas zeroed out):
CROP: https://www.flickr.com/photos/33448355@N07/23840254566/in/album-72157662022611280/
original full pano: https://www.flickr.com/photos/33448355@N07/23783659131/in/dateposted-public/
ISO40, iPhone 6s+ NOT hacked (factory settings):
CROP: https://www.flickr.com/photos/33448355@N07/23570632890/in/album-72157662022611280/
original full pano: https://www.flickr.com/photos/33448355@N07/23866186365/in/dateposted-public/
The differences between the hacked and the original version speak for themselves. The hacked version, while shot at a higher ISO (64 vs. 40), have far much more subtle details. In the non-hacked pano, the lawn not directly lit by the Sun is almost completely "watercolored" (smeared), unlike with the hacked version. There's a dramatic difference.
Color noise and reducing it in post processing
And, even at ISO 64, the additional color noise is pretty low, while the luminance noise is non-existant. Let me also show you a post-processed (in Capture One Pro) hacked image with CNR=80 and LNR=0:
CROP: https://www.flickr.com/photos/33448355@N07/23239783693/in/album-72157662022611280/
Original full pano: https://www.flickr.com/photos/33448355@N07/23498676899/in/album-72157662022611280/
As you can see, the almost entirely color noise-free C1 output is far more detailed than the default (non-hacked) iPhone output. The latter has extensive low-contrast detail smearing; my version, after the CNR applied in C1, has still far less. That is, you'll always get better output if you shoot with my hacks and, if you feel you do need to reduce color noise, post process the results.
All in all,
you make a HUGE mistake if you do want to shoot quality(!) panos, have a jailbroken (or jailbreakable) iPhone 6s+ but don't apply my hack.
The scene in High Quality
For comparison to a decent pano shot with a Fuji X-E1 camera, here's a high-quality pano of almost exactly the same scene (I've shot it some 2-3 meters closer to the statue back in last Summer): https://photosynth.net/view.aspx?cid=2daf1f56-9591-4704-b7ee-f93a959c5883
EDIT (some hours later): I've added another pair of well-lit panoramas with tons of low-contrast foliage, both well-lit and in shadow - now, as opposed to the previous pair, shot at the same ISO 40 sensitivity. Let's start with the two crops:
Hacked: https://www.flickr.com/photos/33448355@N07/23503533929/in/album-72157662552219086/
Original (non-hacked): https://www.flickr.com/photos/33448355@N07/23243277934/in/album-72157662552219086/
The original, full panos: hacked: https://www.flickr.com/photos/33448355@N07/23243271754/in/album-72157662552219086/ ; non-hacked: https://www.flickr.com/photos/33448355@N07/23575751140/in/album-72157662552219086/
Pay special attention to the darker regions of the crop. On the hacked shot, with only a very little (almost invisible!) color noise penalty, the low-contrast details in the shadow areas have far more pronounced details than the in the, because of the heavy-handed noise reduction, smeared original one.
Again, there is very little color noise in the hacked panorama. That's because both shots are shot at the very low ISO 40, there being plenty of light (the brightness value is around 12). While in low light panos (like my two church panos in the previous posts at https://forums.macrumors.com/thread...ne6-noise-filter.1928340/page-2#post-22368606 and https://forums.macrumors.com/thread...ne6-noise-filter.1928340/page-2#post-22368964 ) there were plenty of noise, at low ISO's (bright subjects), there won't be, as is also shown by this example. There, the heavy-handed noise reduction turns out to be absolutely unnecessary.
Note that the previous (Frederik Pacius statue) pano, there was a bit more color noise in the hacked shot. That's because it was shot at ISO 64 (a higher sensitivity), not at ISO 40.
All in all, Apple should have fine-tuned their noise reduction and should have dialed down its strength in low-ISO panoramas like this. Too bad it's only by jailbreaking that you can get significantly better-quality panoramas.
(Note that the two panoramas use visibly different white balance. This has no effect on the results.)
Other, related high-quality panoramas of mine
Should you want to take a look at high-quality panos I've shot of (almost) the same area, check out these:
Hakaniemi beach, to the West from the sauna:
- in daylight: https://photosynth.net/view/8596ccc0-d41f-4c03-adca-a01493b2ab97
- just at the end of the Blue Hour, moonlit (VERY nice!): https://photosynth.net/view/f3bddc7a-3c47-41ff-b741-f0925260f7b7
The Hakaniemi Bus station: https://photosynth.net/view/78475ca7-cb7d-4ed5-96c3-9864a1e282fe
The new panorama has been shot exactly halfway between these two places.
EDIT (2015/12/22 18:32EET): Let me present you another example of the difference. Two ISO 64 (that is, not base but still low ISO) panoramas, one shot using default parameters (non-hacked), and one with a hacked iPhone 6s+. The place is in front of the World Peace statue in Helsinki, Finland. The foliage rendering quality difference in the low-contrast shadow area crops (presented first) speaks for itself.
Crops:
Non-hacked: https://www.flickr.com/photos/33448355@N07/23613997010/in/album-72157662509032902/
Hacked: https://www.flickr.com/photos/33448355@N07/23801385172/in/album-72157662509032902/
Full panos:
Non-hacked: https://www.flickr.com/photos/33448355@N07/23282915253/in/album-72157662509032902/
Hacked: https://www.flickr.com/photos/33448355@N07/23909735075/in/album-72157662509032902/
As with the first (Frederik Pacius statue) non-low-light pano example, there is visible color noise in the hacked shot (after all, this pano wasn't shot at the lowest ISO) but cleaning it with any decent noise reducer app (for example, C1) would have resulted in a significantly better result than the out-of-camera (nonhacked) shot.
The same scene in another high-quality pano I've previously shot: now an early August high-quality shot with decent equipment (sorry, I couldn't auto-stitch the clouds at the top): https://photosynth.net/view/f583d7f4-37f2-42ce-bfd0-9d33d32c90b9 . Don't forget you can browse all my Finnish panoramas at https://photosynth.net/userprofilepage.aspx?user=excellenceinpanorama - they're all high-quality!
EDIT (later the same day): Let me also move a bit towards the North-West, to the southwest corner of the Hakaniemi Market. There, I've shot two series in the unhacked and, then, the hacked mode. Between the two series, I've walked some metres to the West. All four shots are ISO 64 (that is, there being plenty of light, just a bit above base ISO) and present exactly the same smearing of shadow areas as all the other examples presented so far.
In the crops, pay special attention to the plants on the racks. In the hacked images, it can clearly be seen they're differently colored (dark pinkish and green). In the original ones, this can't be seen thanks to both the excessive smearing and color saturation reduction.
Again, these shots being ISO 64 (not base ISO), there's some color noise in the shadows but, as my earlier Capture One noise reduction results have already proved, you can get significantly better image quality out of the hacked originals properly noise filtered in a desktop app like C1 than out of the unhacked iPhone. With the latter, you'll in no way be able to restore the already-lost colors (like the different colors of the plants in this shot) or details. This is why it's highly advantageous to use my hack in good light (too) for the best possible results. (Nevertheless, not even my hack can make the iPhone 6s+ shoot as good panos as even the, today, already fairly inexpensive X-E1 + 27mm combo (some of these panos I've already shown you throughout my post series), let alone panoramas stitched from shots with much less field-of-view and/or shot with full frame / medium format cameras. Assuming proper lens, pano shooting technique and post processing, of course. Quick sweep panoramas will always have much-much lower image quality.)
Series 1:
Crops:
non-hacked: https://www.flickr.com/photos/33448355@N07/23806254442/in/album-72157662518246622/
hacked: https://www.flickr.com/photos/33448355@N07/23914567075/in/album-72157662518246622/
Original, full panos:
non-hacked: https://www.flickr.com/photos/33448355@N07/23832123531/in/album-72157662518246622/
hacked: https://www.flickr.com/photos/33448355@N07/23914679675/in/album-72157662518246622/
Series 2:
Crops:
non-hacked: https://www.flickr.com/photos/33448355@N07/23914780235/in/album-72157660360911623/
hacked: https://www.flickr.com/photos/33448355@N07/23288009503/in/album-72157660360911623/
Original, full panos:
non-hacked: https://www.flickr.com/photos/33448355@N07/23914779165/in/album-72157660360911623/
hacked: https://www.flickr.com/photos/33448355@N07/23888758476/in/album-72157660360911623/
EDIT (12/23): New, well-lit examples of the quality difference:
1. Hévíz, to the West from the main entrance of the lake; ISO 40:
Crops:
non-hacked: https://www.flickr.com/photos/33448355@N07/23635428430/in/album-72157662021430429/
hacked: https://www.flickr.com/photos/33448355@N07/23848606431/in/album-72157662021430429/
Originals:
non-hacked: https://www.flickr.com/photos/33448355@N07/23848680841/in/album-72157662021430429/
hacked: https://www.flickr.com/photos/33448355@N07/23931076205/in/album-72157662021430429/
The shadow regions have excess and very ugly smearing in the non-hacked version.
2. Hévíz, to the East from the main entrance of the lake; ISO 100: these shadow-area crops are much less different than the earlier ones; probably because of the comparatively high (100) ISO sensitivity. Nevertheless, the hacked version still has somewhat better quality:
Crops:
non-hacked: https://www.flickr.com/photos/33448355@N07/23823088362/in/album-72157660384524704/
hacked: https://www.flickr.com/photos/33448355@N07/23304582713/in/album-72157660384524704/
Originals:
non-hacked: https://www.flickr.com/photos/33448355@N07/23304665623/in/album-72157660384524704/
hacked: https://www.flickr.com/photos/33448355@N07/23931356875/in/album-72157660384524704/
Of the several comparative images I've shot, let me show you the ones shot in front of the Frederik Pacius statue, Kaisaniemi park, Helsinki, Finland:
ISO64, iPhone 6s+ hacked (both gammas zeroed out):
CROP: https://www.flickr.com/photos/33448355@N07/23840254566/in/album-72157662022611280/
original full pano: https://www.flickr.com/photos/33448355@N07/23783659131/in/dateposted-public/
ISO40, iPhone 6s+ NOT hacked (factory settings):
CROP: https://www.flickr.com/photos/33448355@N07/23570632890/in/album-72157662022611280/
original full pano: https://www.flickr.com/photos/33448355@N07/23866186365/in/dateposted-public/
The differences between the hacked and the original version speak for themselves. The hacked version, while shot at a higher ISO (64 vs. 40), have far much more subtle details. In the non-hacked pano, the lawn not directly lit by the Sun is almost completely "watercolored" (smeared), unlike with the hacked version. There's a dramatic difference.
Color noise and reducing it in post processing
And, even at ISO 64, the additional color noise is pretty low, while the luminance noise is non-existant. Let me also show you a post-processed (in Capture One Pro) hacked image with CNR=80 and LNR=0:
CROP: https://www.flickr.com/photos/33448355@N07/23239783693/in/album-72157662022611280/
Original full pano: https://www.flickr.com/photos/33448355@N07/23498676899/in/album-72157662022611280/
As you can see, the almost entirely color noise-free C1 output is far more detailed than the default (non-hacked) iPhone output. The latter has extensive low-contrast detail smearing; my version, after the CNR applied in C1, has still far less. That is, you'll always get better output if you shoot with my hacks and, if you feel you do need to reduce color noise, post process the results.
All in all,
you make a HUGE mistake if you do want to shoot quality(!) panos, have a jailbroken (or jailbreakable) iPhone 6s+ but don't apply my hack.
The scene in High Quality
For comparison to a decent pano shot with a Fuji X-E1 camera, here's a high-quality pano of almost exactly the same scene (I've shot it some 2-3 meters closer to the statue back in last Summer): https://photosynth.net/view.aspx?cid=2daf1f56-9591-4704-b7ee-f93a959c5883
EDIT (some hours later): I've added another pair of well-lit panoramas with tons of low-contrast foliage, both well-lit and in shadow - now, as opposed to the previous pair, shot at the same ISO 40 sensitivity. Let's start with the two crops:
Hacked: https://www.flickr.com/photos/33448355@N07/23503533929/in/album-72157662552219086/
Original (non-hacked): https://www.flickr.com/photos/33448355@N07/23243277934/in/album-72157662552219086/
The original, full panos: hacked: https://www.flickr.com/photos/33448355@N07/23243271754/in/album-72157662552219086/ ; non-hacked: https://www.flickr.com/photos/33448355@N07/23575751140/in/album-72157662552219086/
Pay special attention to the darker regions of the crop. On the hacked shot, with only a very little (almost invisible!) color noise penalty, the low-contrast details in the shadow areas have far more pronounced details than the in the, because of the heavy-handed noise reduction, smeared original one.
Again, there is very little color noise in the hacked panorama. That's because both shots are shot at the very low ISO 40, there being plenty of light (the brightness value is around 12). While in low light panos (like my two church panos in the previous posts at https://forums.macrumors.com/thread...ne6-noise-filter.1928340/page-2#post-22368606 and https://forums.macrumors.com/thread...ne6-noise-filter.1928340/page-2#post-22368964 ) there were plenty of noise, at low ISO's (bright subjects), there won't be, as is also shown by this example. There, the heavy-handed noise reduction turns out to be absolutely unnecessary.
Note that the previous (Frederik Pacius statue) pano, there was a bit more color noise in the hacked shot. That's because it was shot at ISO 64 (a higher sensitivity), not at ISO 40.
All in all, Apple should have fine-tuned their noise reduction and should have dialed down its strength in low-ISO panoramas like this. Too bad it's only by jailbreaking that you can get significantly better-quality panoramas.
(Note that the two panoramas use visibly different white balance. This has no effect on the results.)
Other, related high-quality panoramas of mine
Should you want to take a look at high-quality panos I've shot of (almost) the same area, check out these:
Hakaniemi beach, to the West from the sauna:
- in daylight: https://photosynth.net/view/8596ccc0-d41f-4c03-adca-a01493b2ab97
- just at the end of the Blue Hour, moonlit (VERY nice!): https://photosynth.net/view/f3bddc7a-3c47-41ff-b741-f0925260f7b7
The Hakaniemi Bus station: https://photosynth.net/view/78475ca7-cb7d-4ed5-96c3-9864a1e282fe
The new panorama has been shot exactly halfway between these two places.
EDIT (2015/12/22 18:32EET): Let me present you another example of the difference. Two ISO 64 (that is, not base but still low ISO) panoramas, one shot using default parameters (non-hacked), and one with a hacked iPhone 6s+. The place is in front of the World Peace statue in Helsinki, Finland. The foliage rendering quality difference in the low-contrast shadow area crops (presented first) speaks for itself.
Crops:
Non-hacked: https://www.flickr.com/photos/33448355@N07/23613997010/in/album-72157662509032902/
Hacked: https://www.flickr.com/photos/33448355@N07/23801385172/in/album-72157662509032902/
Full panos:
Non-hacked: https://www.flickr.com/photos/33448355@N07/23282915253/in/album-72157662509032902/
Hacked: https://www.flickr.com/photos/33448355@N07/23909735075/in/album-72157662509032902/
As with the first (Frederik Pacius statue) non-low-light pano example, there is visible color noise in the hacked shot (after all, this pano wasn't shot at the lowest ISO) but cleaning it with any decent noise reducer app (for example, C1) would have resulted in a significantly better result than the out-of-camera (nonhacked) shot.
The same scene in another high-quality pano I've previously shot: now an early August high-quality shot with decent equipment (sorry, I couldn't auto-stitch the clouds at the top): https://photosynth.net/view/f583d7f4-37f2-42ce-bfd0-9d33d32c90b9 . Don't forget you can browse all my Finnish panoramas at https://photosynth.net/userprofilepage.aspx?user=excellenceinpanorama - they're all high-quality!
EDIT (later the same day): Let me also move a bit towards the North-West, to the southwest corner of the Hakaniemi Market. There, I've shot two series in the unhacked and, then, the hacked mode. Between the two series, I've walked some metres to the West. All four shots are ISO 64 (that is, there being plenty of light, just a bit above base ISO) and present exactly the same smearing of shadow areas as all the other examples presented so far.
In the crops, pay special attention to the plants on the racks. In the hacked images, it can clearly be seen they're differently colored (dark pinkish and green). In the original ones, this can't be seen thanks to both the excessive smearing and color saturation reduction.
Again, these shots being ISO 64 (not base ISO), there's some color noise in the shadows but, as my earlier Capture One noise reduction results have already proved, you can get significantly better image quality out of the hacked originals properly noise filtered in a desktop app like C1 than out of the unhacked iPhone. With the latter, you'll in no way be able to restore the already-lost colors (like the different colors of the plants in this shot) or details. This is why it's highly advantageous to use my hack in good light (too) for the best possible results. (Nevertheless, not even my hack can make the iPhone 6s+ shoot as good panos as even the, today, already fairly inexpensive X-E1 + 27mm combo (some of these panos I've already shown you throughout my post series), let alone panoramas stitched from shots with much less field-of-view and/or shot with full frame / medium format cameras. Assuming proper lens, pano shooting technique and post processing, of course. Quick sweep panoramas will always have much-much lower image quality.)
Series 1:
Crops:
non-hacked: https://www.flickr.com/photos/33448355@N07/23806254442/in/album-72157662518246622/
hacked: https://www.flickr.com/photos/33448355@N07/23914567075/in/album-72157662518246622/
Original, full panos:
non-hacked: https://www.flickr.com/photos/33448355@N07/23832123531/in/album-72157662518246622/
hacked: https://www.flickr.com/photos/33448355@N07/23914679675/in/album-72157662518246622/
Series 2:
Crops:
non-hacked: https://www.flickr.com/photos/33448355@N07/23914780235/in/album-72157660360911623/
hacked: https://www.flickr.com/photos/33448355@N07/23288009503/in/album-72157660360911623/
Original, full panos:
non-hacked: https://www.flickr.com/photos/33448355@N07/23914779165/in/album-72157660360911623/
hacked: https://www.flickr.com/photos/33448355@N07/23888758476/in/album-72157660360911623/
EDIT (12/23): New, well-lit examples of the quality difference:
1. Hévíz, to the West from the main entrance of the lake; ISO 40:
Crops:
non-hacked: https://www.flickr.com/photos/33448355@N07/23635428430/in/album-72157662021430429/
hacked: https://www.flickr.com/photos/33448355@N07/23848606431/in/album-72157662021430429/
Originals:
non-hacked: https://www.flickr.com/photos/33448355@N07/23848680841/in/album-72157662021430429/
hacked: https://www.flickr.com/photos/33448355@N07/23931076205/in/album-72157662021430429/
The shadow regions have excess and very ugly smearing in the non-hacked version.
2. Hévíz, to the East from the main entrance of the lake; ISO 100: these shadow-area crops are much less different than the earlier ones; probably because of the comparatively high (100) ISO sensitivity. Nevertheless, the hacked version still has somewhat better quality:
Crops:
non-hacked: https://www.flickr.com/photos/33448355@N07/23823088362/in/album-72157660384524704/
hacked: https://www.flickr.com/photos/33448355@N07/23304582713/in/album-72157660384524704/
Originals:
non-hacked: https://www.flickr.com/photos/33448355@N07/23304665623/in/album-72157660384524704/
hacked: https://www.flickr.com/photos/33448355@N07/23931356875/in/album-72157660384524704/
Last edited:
Damn the difference is night and day. It's shameful Apple really didn't care about fine tuning, cause it's clear from your shots that it's completely capable of better shots.
This is reminding me that I have never taken a pano shot till now though lol.
This is reminding me that I have never taken a pano shot till now though lol.
As promised, I've purchased and, then, tested the (current version) of Cortex Camera on the following iOS9.0.1/2 devices:
iPad Air 2
iPhone 5
iPhone 6s+
All my indoor, with (even the Air 2!) one hand handheld, low-light quick test shots turned out to be excellent – much-much better than any shot made with the stock Camera app (even on the OIS-equipped 6s+). I could only find a slight and striclty localized alignment error in only one image (of the nine).
For fun, I've also tested the iOS7 / 8 compatible earlier versions on the iPhone 4 and rMini 2, respectively. (I've also tried to run the app on the iOS6.1.6-based iPod touch 4G and iPhone 3GS but it immediately crashed at asking for location access permission. Manually enabling location access in Settings > Privacy didn't help.)
On the iOS 8.1.2-based rMini2 (again, with an older app version), I found no problems.
On the iOS 7.1.2-based iPhone4 (this may be an even older app version than the iOS8-compliant one),
- the edges in all my tests shots became jagged and having pretty bad oversharpening halos
- the field-of-view was severely reduced (the image “zoomed in”). Note that I haven't introduced more camera shake to my iPhone4 tests shots than to other devices; that is, this severe FoV reduction wasn't caused by (the absolutely necessary) difference elimination between subsequent shots.
All in all, at least based on my first experiments, this app is great if you have at least an iOS8 (but preferably iOS9, as the latest versions are iOS9+ only!) device. (Here, I assume the problems listed with the iPhone4 are common with all the other iOS7 models.) And, of course, if you shoot static subjects.
I'll continue posting my findings - stay tuned!
EDIT (next day): the 5 Mpixel (the same Mpixel number as the iPhone4) iPad3 running on iOS8 (that is, only able to run an older version of the app) produced excellent pictures. While I did spot unnatural edge enhancement (and halos), the overall results were far superior to the still shots produced by the camera app, with far-far more detail.
Last edited:
That is still evident in high-contrast scenes (iPhone 6+ iOS 9). That is the main drawback that I have against Cortex Cam. For most normal scenes it is not that noticable, but I've done image processing research for a long time and can usually spot artifacts that normal people don't see...
Otherwise it is an excellent app.
Yup, at the borders of extreme contrast changes (e.g., a backlit tree trunk or building wall edge) there definitely are more pronounced halos than in the stock photos. And, of course, it can't make use of goodies like live photos (on the 6s/6s+) or even exposure compensation.
I still need to test whether the this subpixel matching + frame adding, which increases resolution and decrease noise, also results in DR enlargement. Given that the noise is definitely lower than in stock photos, I'd say it must have as good DR as the semi-two-frame-stitcher built-in HDR of the stock app. Too bad there is no manual exposure compensation, which would definitely help with metering for the highlights. Fortunately, at least (as opposed to the stock Camera app) it supports separate focus and exposure locking / metering, which means one can first lock the focus (by long-tapping the bracket) and, then, tap anywhere on the screen viewfinder to do the exposure metering.
Nevertheless, in all my tests (today, I've continued doing them), apart from these problems, the app produced excellent results on both the iPhone 6s+ and the iPhone 5. It produced appr. as good image as the 16 Mpixel micro 4/3 Panasonic GM1 + 12-32 combo operating at 28mm equiv, f4.5 (sweet spot to maximize sharpness and get rid of optical problems), ISO200 and mechanical shutter. It was only the Nokia 808, operating at its (hacked) 41 Mpixel mode, that produced better-detailed images. In all cases, the Cortex images were better-detailed and less noisy than those of the stock Camera.
Oh nice, I'll try to test the pano out tomorrow! I can't wait for the regular photos to be fixed, my family photos look meh at default.
Off topic: Just realized earlier in the week that I still have your repo installed lol.
I’m not much into coding, but there was a flex 2 patch that claims to improve the stock camera photos. Anyone that has flex2 care to see how it compares to the tweak posted here?
You mean "Upgraded Camera Abilities Front and Back"? I've tested it on the 6s+. It only enables features not available in some pre-iPhone6s models (e.g., HDR), in addition to increasing the max gain (ISO). It doesn't get rid of the NR - that is, it's incapable of improving IQ WRT NR.
Like Menn mentioned it's a totally different tweak and just turns 6s features on for older phones.
Ya Menneisyys2 is is pretty skilled with this stuff. I'm just sad Apple did this and that he had issues getting this to work outside of pano mode.