Koepi
Posts: 68
Joined: Fri Apr 29, 2016 7:18 am

Solution planned for the bad lenses of the V2 8MP RasPi cam?

Sun May 01, 2016 9:46 am

You're going to hate me for this question, but here it comes anyways:

As being discussed and shown here (viewtopic.php?f=43&t=145815) as mass phenomenon, the Pi Cam V2.1 has serious issues producing sharp images. The lens is bad, plain and simple.

For 30 Euros this is quite a dissapointment - that is some real amount of money for me and most likely for others as well. The really great sensor is crippled by the cheapest and worst optics which seemed to be around somewhere. The cheap V1.3 cam (for 13 Euros, for comparison) produces way sharper images.

I'd like to know if there is a solution planned and how it will look like.

anykey
Posts: 16
Joined: Fri Nov 29, 2013 8:11 pm
Location: Netherlands
Contact: Website

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Mon May 02, 2016 7:36 pm

Hi,

I am not sure the lens is really less good then the v1.3 lens, but the default distance of focus is for sure very different. With the v1.3 lens, the sharp area was from about 70cm and beyond.

The new camera the lens is much more "macro" setup. Images from 20cm are already super sharp, objects in the distance are not :-(

Check out some comparable results here:
http://www.pi3dscan.com/index.php/blog2 ... -pi-camera

6by9
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 7026
Joined: Wed Dec 04, 2013 11:27 am
Location: ZZ9 Plural Z Alpha, aka just outside Cambridge.

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Mon May 02, 2016 7:55 pm

Investigations are ongoing.
Software Engineer at Raspberry Pi Trading. Views expressed are still personal views.
I'm not interested in doing contracts for bespoke functionality - please don't ask.

fruit-uk
Posts: 609
Joined: Wed Aug 06, 2014 4:19 pm
Location: Suffolk, UK

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Mon May 02, 2016 7:56 pm

Koepi wrote:I'd like to know if there is a solution planned and how it will look like.
In another thread it was stated on Saturday that discussions are being held.

You may not be aware but today is a Bank Holiday in the UK and a holiday in other parts of Europe so there may not be many people actually at work in the various involved organisations to discuss with until tomorrow at the earliest

Koepi
Posts: 68
Joined: Fri Apr 29, 2016 7:18 am

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Mon May 02, 2016 8:19 pm

@6by9: Thank you for confirmation that the team will look into the issue!


@fruit-uk: I nowhere stated that I "demand immediate response" as you are implying, I just asked if there is a solution planned. No need to rush things, no time frames, just out of interest.

You're very lucky in the UK! In Germany such holidays are simply lost when they are on a Sunday. Never would have imagined that looking at the system Thatcher built/stood for otherwise.

User avatar
Paeryn
Posts: 2614
Joined: Wed Nov 23, 2011 1:10 am
Location: Sheffield, England

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Mon May 02, 2016 9:42 pm

Koepi wrote:You're very lucky in the UK! In Germany such holidays are simply lost when they are on a Sunday. Never would have imagined that looking at the system Thatcher built/stood for otherwise.
Whilst technically May Day is the 1st of May, in the UK the May Day bank holiday is always on the first Monday in May so we can never loose it to a Sunday ;-)
She who travels light — forgot something.

User avatar
jbeale
Posts: 3471
Joined: Tue Nov 22, 2011 11:51 pm
Contact: Website

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Tue May 03, 2016 12:27 am

Koepi wrote: The lens is bad, plain and simple.
That's what I thought at first, and some units may indeed be bad. But I would call my result from one test at 30 cm at least adequate http://bealecorner.org/best/RPi/PiCamV2-test7.jpg and I feel the one at 40 cm from hystrix is really quite good (at least, for the center area where the test pattern is): https://www.dropbox.com/s/dlav27jry1dwze5/new_cam4.jpg Yes, the strong moire color pattern on fine line patterns is an image defect, but it would not be present if the lens had poor resolving power. For those specific units, the problem seems to be simply the focus setting is in the macro range, and not at infinity.

Koepi
Posts: 68
Joined: Fri Apr 29, 2016 7:18 am

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Tue May 03, 2016 4:26 am

When looking at my full resolution images, there are some spots with 'ok' focus. Close to the center, between 10 and more than 50m away (looking at the fine branches of the trees).
Looking at your fine print photo I see the same 'wash out' effect on the letters, that is also not sharp. Its good enough so you can read it, but compare that to a photograph made with your smartphone.
Also, when you swapped lenses you immediately got more-than-decent results. That makes me quite sure that the lens is unsuitable and faulty.

michelpi
Posts: 12
Joined: Fri Feb 06, 2015 8:34 pm

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Tue May 03, 2016 6:00 am

For the manufacturer, the solution can't be that hard. I think the sensor is fine, they just need to adjust the lens position by 20-30 micrometers.

If you don't want to mess with the focus yourself, I would just return them to the shop and ask for a refund. At some point someone will make a focus-able v2 camera, and you get yourself one of those.

User avatar
jbeale
Posts: 3471
Joined: Tue Nov 22, 2011 11:51 pm
Contact: Website

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Tue May 03, 2016 5:07 pm

Koepi wrote:When looking at my full resolution images, there are some spots with 'ok' focus. Close to the center, between 10 and more than 50m away (looking at the fine branches of the trees).
Looking at your fine print photo I see the same 'wash out' effect on the letters, that is also not sharp. Its good enough so you can read it, but compare that to a photograph made with your smartphone.
Also, when you swapped lenses you immediately got more-than-decent results. That makes me quite sure that the lens is unsuitable and faulty.
I'm not as sure. I agree there is a "soft-focus" effect on the fine print example, but that high-contrast target on uniform white background makes this effect more apparent than the backyard scene, so the v2.1 board with v1.3 lens may have it as well. I would need both configurations recording the same target to compare apples-to-apples. The soft focus effect can be modelled as a sum of a sharp and defocused image, and can be reduced by post-processing steps (eg. unsharp mask). I believe noise reduction is also being done which can have similar effects on fine detail. All these sensor systems do that type of processing, the difference is how much (a tuning parameter, and personal judgement of whoever does the tuning). Anyway at this point it's clear there's a problem to address and I doubt I'll spend more time on experiments. I will wait for further announcements.

User avatar
MarkHaysHarris777
Posts: 1820
Joined: Mon Mar 23, 2015 7:39 am
Location: Rochester, MN
Contact: Website

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Tue May 03, 2016 7:11 pm

jbeale wrote:Anyway at this point it's clear there's a problem to address and I doubt I'll spend more time on experiments. I will wait for further announcements.
Uffda. Are you recommending the community hold-off on purchasing the V2 camera until things get sorted out...

... this reminds me a bit of our now infamous Hubble Telescope with the defective mirror... we have an 8megapixel camera with a lens issue... sheesh.
marcus
:ugeek:

User avatar
jbeale
Posts: 3471
Joined: Tue Nov 22, 2011 11:51 pm
Contact: Website

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Tue May 03, 2016 7:59 pm

viewtopic.php?f=43&t=146579#p966456

I'm not making "recommendations for the community." I do presume at some point the above-referenced ongoing investigations will reach some conclusion.

User avatar
MarkHaysHarris777
Posts: 1820
Joined: Mon Mar 23, 2015 7:39 am
Location: Rochester, MN
Contact: Website

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Tue May 03, 2016 8:08 pm

jbeale wrote:viewtopic.php?f=43&t=146579#p966456

I'm not making "recommendations for the community." I do presume at some point the above-referenced ongoing investigations will reach some conclusion.
Thanks. heh, I have read the entire thread; was all set to purchase the new camera but may wait now.

I really think the lens should be adjustable; honestly. The lens should not be fixed (not glued) ; it should be tight (snug) and the user should set it according to macro or infinite (whatever).
marcus
:ugeek:

User avatar
jbeale
Posts: 3471
Joined: Tue Nov 22, 2011 11:51 pm
Contact: Website

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Tue May 03, 2016 9:20 pm

MarkHaysHarris777 wrote:I really think the lens should be adjustable; honestly. The lens should not be fixed (not glued) ; it should be tight (snug) and the user should set it according to macro or infinite (whatever).
That makes perfect sense to me, for a RPi camera. Problem is (I think) the RPi sort of application did not exist until very recently; the volume* use cases the manufacturers designed for was mobile devices, and maybe car backup and dash cams. Those were supposed to be simple, sealed, no-adjustment consumer items. So they only provided for loose fitting lenses, one-time factory-set which need glue to keep them in place. All this just my personal guess. If you did make the existing lens a tight-fit but adjustable, you would likely still need a tool for the adjustment, as the lens is so small (6mm size).

*several billion CMOS sensors per year(!) http://electroiq.com/blog/2011/01/wafer ... e-sensors/

User avatar
MarkHaysHarris777
Posts: 1820
Joined: Mon Mar 23, 2015 7:39 am
Location: Rochester, MN
Contact: Website

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Tue May 03, 2016 11:15 pm

jbeale wrote:
MarkHaysHarris777 wrote:I really think the lens should be adjustable; honestly. The lens should not be fixed (not glued) ; it should be tight (snug) and the user should set it according to macro or infinite (whatever).
(main use case)... Those were supposed to be simple, sealed, no-adjustment consumer items. So they only provided for loose fitting lenses, one-time factory-set which need glue to keep them in place ...
Yes. Maybe the foundation needs to design their own rather than vend it out... :?

The other idea I had was a rotating ocular ... similar to the lower stage of a common microscope; infinite, portrait, and macro (rotate the lens of choice into place).
marcus
:ugeek:

User avatar
Gavinmc42
Posts: 3457
Joined: Wed Aug 28, 2013 3:31 am

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Wed May 04, 2016 12:59 am

Rotating optics like this?
https://www.youtube.com/watch?v=0oek_-0jbX0

Mobile phone clip on?
Two telephoto/zooms, wide angle and macro.

Nice scale 3D printed case with Pi inside.
http://www.tvcameramuseum.org/
RCA, Marconi, EMI, Pye

Still not convinced it is all focus issues, not until I make a lens twister and decent solid mount.
Night shots have no noise on normal settings, it could be the noise rejection software.

Lots more tests to get some answers, it could just be a simple software fix.
Or a very complex software fix:)

In the meantime I'm glad I got a bunch of V1's when they went on special.
The Raspberry Pi is all about learning, now we are learning about image sensors, optics and camera software.
I am having fun:)

For all the whingers.
20 years ago I dreamed about being able to do this stuff, camera chips were hard to get.
Real camera buffs get only raw data and process it themselves anyway.
And they have a bunch of lenses with corrections for each one.
I can do this too now for a fraction of the cost a "real" camera.
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

User avatar
MarkHaysHarris777
Posts: 1820
Joined: Mon Mar 23, 2015 7:39 am
Location: Rochester, MN
Contact: Website

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Wed May 04, 2016 1:48 am

Gavinmc42 wrote:Rotating optics like this?
https://www.youtube.com/watch?v=0oek_-0jbX0
Yes, precisely... just tinier. ;)
marcus
:ugeek:

User avatar
jbeale
Posts: 3471
Joined: Tue Nov 22, 2011 11:51 pm
Contact: Website

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Wed May 04, 2016 3:41 am

Koepi wrote:Looking at your fine print photo I see the same 'wash out' effect on the letters, that is also not sharp. Its good enough so you can read it, but compare that to a photograph made with your smartphone.
Since you suggested that, I tried that. My phone is a Samsung GT-I8190 with only 5 MP camera. See what you think, full images:
8 MP RPi camera: http://www.bealecorner.org/best/RPi/PiCamV2-test7.jpg
5 MP Samsung cam: http://www.bealecorner.org/best/RPi/Sam ... 202838.jpg

...and here are center crops: from RPi v2.1, then Samsung:
ImageImage
This is far from a careful test; feel free to do you own. Target is the instruction leaflet that came in the v2.1 camera box, spread out and taped down flat.

EDIT: The RPi camera was 30 cm away from the target for that shot, chosen because that is approximately the best focus point on my particular camera. Samsung phone camera was at approximately the same distance, chosen to match the framing. A few other RPi v2 photos at the bottom of this page: http://www.bealecorner.org/best/RPi/index.html#RP2 including a RAW+JPEG file example, which with some detective work you could extract the raw data and assemble.
Last edited by jbeale on Wed May 04, 2016 4:07 pm, edited 5 times in total.

User avatar
Gavinmc42
Posts: 3457
Joined: Wed Aug 28, 2013 3:31 am

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Wed May 04, 2016 4:38 am

8M is better than the 5M?
Or did you get the focus sweet spot?
V2 is focused at 50cm?
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

racer993
Posts: 60
Joined: Mon Feb 18, 2013 7:27 pm

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Fri May 06, 2016 7:52 am

Has anyone seen a formal statement to the issue? We have used about 1.500 RPi OV5647 camera modules in the past 6 months and would like to continue to use the RPi + camera board. I would, really like to know whether a solution is at hand.

Focus from 50cm - infinity would be great.

User avatar
jbeale
Posts: 3471
Joined: Tue Nov 22, 2011 11:51 pm
Contact: Website

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Fri May 06, 2016 1:58 pm

So far I have seen that "discussions are being had" (30 April) and "investigations are ongoing" (2 May).
viewtopic.php?f=43&t=145815&start=100#p964982
viewtopic.php?f=43&t=146579#p966456

Here are some examples, it is particularly noticeable for me with distant objects in the corner of the frame. If you aren't too critical of image quality, and don't look at the corners of the frame, some people might not notice a problem.
viewtopic.php?f=43&t=147109#p968457
viewtopic.php?f=43&t=147109#p968620

To achieve "focus" from 50 cm - infinity with f=3mm, f/2.0 lens, you must accept a Circle of Confusion of 4 um, which on this sensor is about 4 pixels- not very sharp! If you demand a Circle of Confusion of 1 pixel (~ 0.001 mm) and focus at 3 meters, theory says you can only get 2 - 9 m into focus. At a more realistic CoC= 2 pixels, you can get 1 m - infinity. Have a play with http://www.dofmaster.com/dofjs.html
Last edited by jbeale on Fri May 06, 2016 8:17 pm, edited 2 times in total.

racer993
Posts: 60
Joined: Mon Feb 18, 2013 7:27 pm

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Fri May 06, 2016 3:17 pm

thx for the explanation jbeale.

The current lens quality is not good enough for our purpose, it is a big no-go.

A drop-in replacement, as advertised, would be a minimum requirement, including a lens with a fov similar to the Raspberry Pi Camera version 1.3.

I would even favor a smaller field of view in order to get higher detail rather than a bigger picture.

Hopefully it won't be to long until the official investigation can be concluded.

User avatar
jbeale
Posts: 3471
Joined: Tue Nov 22, 2011 11:51 pm
Contact: Website

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Fri May 06, 2016 8:29 pm

I agree. I would expect anyone who pays attention to image quality would be critical of the v2 camera when used on distant subjects; at least for the units I received. I suppose suppliers might not be hasty with statements that have financial implications, given that not everyone seems to have noticed a focus issue on v2.

EDIT: did a few more tests with my Pi NoIR v2 board. On this one, the lens not bad for objects 3 m away in the center, but the corners of the frame are severely blurry. The focal distance at the corners of the frame on this module seems to be near 30 cm. This is different from the normal Pi v2 board I tested first, where the entire field seemed to be near focus at 30 cm. So the two boards are not consistent, which is perhaps worse than being consistently off in the same way.

User avatar
jbeale
Posts: 3471
Joined: Tue Nov 22, 2011 11:51 pm
Contact: Website

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Sat May 07, 2016 5:12 am

Question: would you get the effect I see on my NoIR v2 board, if you used a lens designed for the extra optical path length of an IR-cut filter, and then simply removed that filter without replacing it with clear glass of the same numerical index, and then adjusted the lens for best focus at the center? If the (missing) filter had significant thickness (longer path length at the corners with steeper angle of incidence) then I think removing it could have this close-focus-at-the-corners effect. Just a thought.

Edit: No, can't be. Sign is wrong, close focus in corners means extra optical path length behind the lens at the corner.

In this outdoor shot from Pi NoIR v2 linked below, you can see the roof tiles at the center of the frame (maybe 15 m away) are indeed in focus, although some might prefer a bit more sharpening in the ISP. Meanwhile the corners of the frame are clearly OOF.
http://bealecorner.org/best/RPi/PiNoIRv2-outside.jpg

EDIT: Here's another test showing the difference between v1.3 and v2.1: viewtopic.php?f=43&t=145815&start=175#p970337

User avatar
Mettauk
Posts: 234
Joined: Mon Dec 10, 2012 12:40 pm
Location: Zarg

Re: Solution planned for the bad lenses of the V2 8MP RasPi

Mon May 09, 2016 9:49 am

On Mon May 02, 2016 7:55 pm
6by9 wrote:Investigations are ongoing.
Any official news before I return 4 of the new V2 cameras i purchased?

I will keep one of each and attempt a physical hack of the lenses to refocus them at infinity but I'm not up for this on all of them, will stick with V1.3 or often superior clones!
As humans we have been the same for a very very long time, technology changes how we do... not who we are as people.

Return to “Camera board”