How to make flats for very wide field imaging?

[marekintheuk]

I'm struggling to make good flats for a 10mm focal length lens (on a DSLR).

I've tried back-illuminating various materials which make getting even illumination relatively easy, but I have failed to find a material with sufficiently even translucency not to cause mottling in the flat.

Reflectance of materials seems better (more even) but then I have to problem of getting even illumination on a large surface close to the lens.

I have an off-the-shelf  profile for the lens (Sigma 10-20 f.3.5) but unfortunately it does not include vignetting data.

My last resort is to try and do it piece-wise - i.e., create a small, distant illuminated target and move it around the lens field of view. Plausible but hard work.

Any other suggestions?

[pipnina]

Something like this might provide a good surface to shoot at for flats? https://www.amazon.co.uk/32-Inch-Portable-Translucent-Collapsible-Reflector/dp/B002ZIVKAE/ref=sr_1_1_sspa?ie=UTF8&qid=1546970257&sr=8-1-spons&keywords=photographic+reflector&psc=1

Also, flats don't need a perfectly reflective surface because your lens should be focused at infinity while the target is considerably closer.

[marekintheuk]

The problem is that the lens has a very wide angle of view so the target needs to be within a couple on inches of the lens (unless it's massive). This makes getting even front-illumination very difficult. Also since the depth of field is so large, the target is only just out of focus, so any unevenness shows up in the flat. 

[happy-kat]

I use the android app lightbox and run that on my tablet and sit it on the lens, camera in manual focus and same iso and aperture, switch it to AV mode which sorts the exposure time.

[marekintheuk]

5 minutes ago, happy-kat said:

I use the android app lightbox and run that on my tablet and sit it on the lens, camera in manual focus and same iso and aperture, switch it to AV mode which sorts the exposure time.

Have you tried this for a wide angle lens though? A typical tablet display does not produce constant light output with viewing angle. It OK for long focal lengths but not for 10mm where there 45 degrees difference from the center to the edge. 

[happy-kat]

My Nexus 7 illuminates pretty well and evenly if you have a suitable tablet then try it the app is free and on ios too.

[marekintheuk]

10 minutes ago, happy-kat said:

My Nexus 7 illuminates pretty well and evenly if you have a suitable tablet then try it the app is free and on ios too.

Just did. There's about 20% less light at 45 degrees (edge of FoV) than 90 degrees (center of FoV) . Not really good enough for a decent flat. 

[happy-kat]

I guess this is because your lens is I suspect convex on the front element rather than flat.

[Scooot]

I use a photo of a white piece of Paper on my IPad. I put a white cloth over the lens. So you could also use a similar method with a large flat screen TV. I can view my Images stored on my iPad on the TV so I could also view the Image of my white piece of paper on the TV. The only snag is the camera wouldn’t be at the same orientation as the lights, but you could ensure everything else is the same. I’d guess the screen would be large enough to fully illuminate the 10mm lens. It’s an idea anyway?

Edited January 8, 2019 by Scooot

[marekintheuk]

2 hours ago, happy-kat said:

I guess this is because your lens is I suspect convex on the front element rather than flat.

Hmm, not sure why the lens shape should make any difference. Besides, the fall-off test I did was with a long focal length lens - I just rotated the tablet to measure the dispersion pattern. Interestingly enough, I also noticed that the tablet dispersion pattern was different in the vertical and horizontal direction which complicates matters.

[marekintheuk]

1 hour ago, Scooot said:

I use a photo of a white piece of Paper on my IPad. I put a white cloth over the lens. So you could also use a similar method with a large flat screen TV. I can view my Images stored on my iPad on the TV so I could also view the Image of my white piece of paper on the TV. The only snag is the camera wouldn’t be at the same orientation as the lights, but you could ensure everything else is the same. I’d guess the screen would be large enough to fully illuminate the 10mm lens. It’s an idea anyway?

I've tried various materials, but since anything in front of the lens is almost in focus, any unevenness in the density of the material shows up in the flat. All these ideas are fine at long focal lengths, but not at 10mm! I believe that in the serious optical measurement world they use something called an 'integrating sphere' but that's not something you can lash up in a garage!

Looks like I'm going to have to do it the hard (piece-wise) way.

 

[michael8554]

If you're convinced that tablet and laptop screens don't give even illumination at different angles, then you should try sky flats, or a light box:

https://www.ebay.co.uk/itm/A4-Drawing-Light-Board-for-Diamond-Painting-Kits-Tracing-Light-Pad-Box-Table-M6/223224674759?hash=item33f93a39c7:g:CUwAAOSw1cdb5vrm

Michael

[Scooot]

8 hours ago, marekintheuk said:

I've tried various materials, but since anything in front of the lens is almost in focus, any unevenness in the density of the material shows up in the flat. All these ideas are fine at long focal lengths, but not at 10mm! I believe that in the serious optical measurement world they use something called an 'integrating sphere' but that's not something you can lash up in a garage!

Looks like I'm going to have to do it the hard (piece-wise) way.

 

I’d be surprised a piece of white cloth attached to the front of the lens with an elastic band would be in focus, and anything beyond it wouldn’t be either?

Edited January 9, 2019 by Scooot

[happy-kat]

Fucus should be the same as when taking the lights.

[happy-kat]

Thinking about it you want that light fall off as that is what your lens sees so that is what the flat needs to see, sounds like the 11mm lens probably shows a fair amount of vignetting.

[marekintheuk]

7 hours ago, Scooot said:

I’d be surprised a piece of white cloth attached to the front of the lens with an elastic band would be in focus, and anything beyond it wouldn’t be either?

OK, the first picture is of a 2cm 'H' drawn on some paper and held right against the lens (no hood) and using a LCD monitor as a lightbox. Perhaps not 'in focus', but the shape is clear and you can easily see the texture of the paper.

The second and third picture show what I get with the lens right up against the monitor (no paper). The difference is just that the camera was rotated from landscape to portrait between them to show the asymmetry of the light dispersion from the screen.

The final picture is the best I can currently do which is of an indirectly illuminated ceiling. It shows a more realistic extent of the lens vignetting (compared this to the exaggerated fall-off when using a screen). It still has too much texture and unevenness to be used as a flat, but it does show how much worse the other approaches are.

All images are linear TIFFs convert to JPG and downscale to 10%.

 

[marekintheuk]

15 hours ago, michael8554 said:

If you're convinced that tablet and laptop screens don't give even illumination at different angles, then you should try sky flats, or a light box:

https://www.ebay.co.uk/itm/A4-Drawing-Light-Board-for-Diamond-Painting-Kits-Tracing-Light-Pad-Box-Table-M6/223224674759?hash=item33f93a39c7:g:CUwAAOSw1cdb5vrm

Michael

I've posted some pics further up the thread showing how poor screens are (for a wide-angle lens). Sky flats don't work either for a lens with a 90 degree FoV - especially in the cloudy UK.I haven't tried a lightbox like this, but I doubt the it provides sufficiently even illumination. It probably has a array of LEDs (perhaps a dozen) under some translucent perspex - fine for its intended purpose, but I bet if I use it to make flat and then stretch the image I'll be able to count the LEDs.

Edited January 9, 2019 by marekintheuk
'up' the thread instead of 'down'...

[Davey-T]

You could try making a custom light box to suit though as you say the wide angle a depth of field makes it tricky.
I made one for my SCT out of white plastic material illuminated by halogen bulbs shielded to give diffuse light with dimmer switch.

There was a recent post of a similar one.

Dave

 

 

[RayD]

Are you not talking yourself in to the conclusion that this wide an angle lens is not suitable for AP when flats are needed?  

I can only think of something like a custom made flat box, but with diffused lighting from behind the front face of the lens.  The diffused and reflected light would mean an even illumination, and the rear orientation of the light source means you won't see LED's.

[RayD]

1 minute ago, Davey-T said:

You could try making a custom light box to suit though as you say the wide angle a depth of field makes it tricky.
I made one for my SCT out of white plastic material illuminated by halogen bulbs shielded to give diffuse light with dimmer switch.

There was a recent post of a similar one.

Dave

 

 

How spooky that I was writing the same thing at the same time!! 

[marekintheuk]

22 minutes ago, RayD said:

Are you not talking yourself in to the conclusion that this wide an angle lens is not suitable for AP when flats are needed?  

I can only think of something like a custom made flat box, but with diffused lighting from behind the front face of the lens.  The diffused and reflected light would mean an even illumination, and the rear orientation of the light source means you won't see LED's.

That's defeatist!

The front-illuminated lightbox approach is possible, but it's not obvious how to get even illumination (to the required precision). I suppose one would be make the illumination as even as possible and then measure the variation with a spot meter and use that as a correction function.

Actually I have a couple of ideas that might work:

(1) Stick camera+lens on a pano-head. Take a series of 2D heavily overlapping images. Scene should have plenty of detail and static illumination (probably indoors, hence pano-head to ensure rotation about nodal point). Load up picture into Hugin/panotools and let the photometric optimiser work out the vignetting it needs to correct to stitch the images. Extract polynomial parameters and convert to lens profile or plot a flat. In fact I usually use Hugin to remap my wide-field images to a fisheye projection, so the photometric correction can stay in Hugin as part of the pre-stacking workflow.

(2) Stick camera+lens on a pano-head. Set up a light source and take a series of pictures with the light source at different points in the FoV (in effect a 'serialised' integration sphere). Measure brightness of source in images and fit a polynomial to the data points (I already have code to do that for background extraction). Convert polynomial to profile or flat as above.

But I was hoping someone might have a simpler method (that works).

 

 

[RayD]

3 minutes ago, marekintheuk said:

But I was hoping someone might have a simpler method (that works).

 

Hmmm, well I wish you luck with this.  I wouldn't think there are too many people shooting 90 degree FOV subs other than with an all-sky camera, as do I, but I don't apply flats to the images.

[marekintheuk]

On 08/01/2019 at 20:49, happy-kat said:

I guess this is because your lens is I suspect convex on the front element rather than flat.

I had a revisit of this suggestion over the weekend. There is certainly a suggestion on the lensfun lens calibration mailing list that non-constant gap between the diffuser and the surface of the front element can distort the vignette calibration, so I did some tests. I varied the diffuser-element spacing from 0cm to 9cm. The front element curvature is about 5mm deep, so that should have made a discernible difference. Result: No difference above the noise floor.

So I went back to plans A (piecewise data by rotating the camera with a fixed light source) and B (using Hugin to estimate the vignetting correction polynomial from a series of overlapping-shifting images. I consider the piecewise data to be the 'gold standard' in terms of vignetting since it is by definition free from the angular dispersion problems - although obviously too sparse and noisy to be anything other than test data.

For comparison I also took a series of calibration images (flats) using (a) the method recommend by lensfun (a translucent diffuser pointing up at an evenly lit ceiling), (b) the with same diffuser backlit by a tablet and (c) just the tablet. The result are shown below (16b luminance value plotted against pixels-from-left-edge).

1. The 'poly' line is Hugin's estimated polynomial from the piecewise data plotted to give a synthetic flat. Not bad, low noise obviously, but deviates from the data in the centre of the image. 

2. The noisy red 'd0so' and blue 'd9cmso' lines are from the lensfun lightbox method (0-9cm lens-diffuser separation). They are clearly the right shape *BUT* in order to get them to fit I had to offset the black point significantly (+2500 for d0so and +4200 for d9cmso).

3. The smooth green line is from the tablet-plus-diffuser with some wavelet noise reduction applied. Very similar to the lensfun method and a good fit to the data (with black point adjustment again).

4. The mauve 'tabso' line is from using the tablet without a diffuser. It is pretty clear that there is no way that this curve can fit the piecewise data (linearly).

The diffuser in the above cases was a nice find (I think): multiple sheets of premium photo paper (Canon Platinum Pro) - no logos and very internal little texture. I used 2 sheets - 1 didn't provide quite enough diffusion, perhaps more would be better but then noise becomes more of an issue (but I can always stack my way round that).

So I think I have a good method of producing a wide-field (>90 degree) flat. I'm not sure I understand the source of the 'black point offset' required when using the diffuser - I'll have to think about that a bit more - but empirically I have a method that works. Hopefully this might be useful to someone else one day too!

 

[PhilRosenberg]

I just wanted to thank you for posting your comparisons of your attempts. I have been looking for a good way to do this for a long time. The diffuser over a screen method is something I hadn't thought of. I'm going to order a piece of translucent white perspex and give this a try.

Thanks again

Phil

Edited November 28, 2022 by PhilRosenberg

[Phil_rosenberg]

Haven't tried yet, but the A3 sized sheet of 3mm white perspex arrived today. I'm quite hopeful that sat in front of my laptop screen it will give a good flat 🤞