DSLR Flat Frames Light Box

[TiTanecd]

Hey there

After my first wide view picture I took on Sunday, while post processing I noticed the Vignetting effect on the corners. I know its down to Flats. I haven't used them yet because I haven't got to the stage of experience, and its good I feel to learn like this. Instead of going all out.

I used my 500D only with the stock 18-55mm lens (at 18mm) and thought how can I get flats from this. I know you can use a tshirt over the OTA but stretch something over the lens 1. is harder 2. will mess up the focus. 

Google picture led me to this guy that has a light box in his oby

Can I use a similar setup for the DSLR, have a well evenly lit box that sits infront of the lens? Main reason for asking is 'can there be distand between the light and lens'? 

Tx guys

[Bizibilder]

Yes - but it can be easier to make the box sit directly on the end of the scope.  You can also do as in the picture but have a sheet of white "stuff" illuminated from the same side as the scope ie reflecting off the white "stuff".  Or you can use a white laptop screen (if you can find one big enough) or even one of these: http://www.ebay.co.uk/itm/300-300mm-Ultra-Thin-LED-Ceiling-Flat-Panel-Light-Downlight-Bulb-Cool-White-/390877102464?pt=UK_HomeGarden_Lighting_Lamps_Lighting_SM&var=&hash=item5b02174180

(Usual disclaimer)

[TiTanecd]

Tx for the clarification, was thinking of doing something like the ceiling light. Have LEDs behind a white perspex pane. 

[IanL]

The main problem with making a lightbox for a DLSR (or any OSC camera) is that you need a true white light source for best results.  With a monochrome camera you can vary the exposure length to ensure that flats for each filter reach the same brightness, but with a DLSR/OSC you can only take one exposure and you may end up with a colour cast in the flat frames if the source puts out different amounts of light at different wavelengths.

A colour cast is not a problem in itself, as you can re-balance the colours easily enough when processing the image.  The problem is that you will (typically) have least noise in the green channel, more in the blue and a lot more in the red.  If the difference in noise levels is small it's not a problem, but with all my self-build attempts I ended up with an incredibly noisy red-channel which in turn adds noticeable noise to the final image.

Typically you will make a diffuser from one or more layers of white acrylic with a gap between each layer (the rough/diffusing type is better than smooth/shiny).  You can pick it up in various sizes on eBay or try your local sign-makers for offcuts.

- The best light source is a cloudy daylight sky, since this gives an even illumination, and you can just make a diffuser out of a bit of acrylic with a bit of foamboard or light plywood to hold it over the scope.  You need to seal around the end of the OTA so that no light leaks in from the edges.

- You can also use a diffuser on a sunny day provided the kit is not in direct sunlight, but you need to take more care to diffuse the light (two or more layers of acrylic in box that is light-proof except the end that you point at the sky for example).

- If you want to make a light box to use at night you need to go a bit further - the usual method is to have a bunch of lights shine up in a box and reflect of the inside-top of a closed (white) box back down on to a double-layer of diffuser sitting on the OTA.  The light sources need to be shielded from directly shining on the diffuser surface.  They're fairly easy to put together using foamboard and hot glue.

- The trick is getting a good light source.  Old fashioned incandescent lamps are best as they give a nice broad spectrum and no major problems with a colour cast - e.g. torch bulbs or low voltage garden lights but they're getting harder to come by with the rise of LEDs for this kind of thing.

- LEDs are much harder - I tried three different methods for making an LED light box and all had the same problem of a low level of red and a lot of noise in that channel.  LEDs do not give out a continuous spectrum, only very narrow wavelengths.  You can try using red, green and blue LEDs to make a white light source but it is very hard to find sufficiently bright red ones compared to the blue and green - even adjusting the output using resistors I could not do it (as the greed/blue will stop emitting if the input is too low).  White LEDs are not really white at all, even though they look to be to they human eye.  Again they typically seem much weaker in the red channel.

- Lots of people use EL panels but again you need to be careful because some types will give a colour cast even if they appear white to the eye.  There are suppliers of ready made panels (e.g Gerd Neumann) that have been tested to give a broad spectrum, but a lot of people complain about all EL panels having very flimsy connections between the wires and the EL foil itself which tend to break at the slightest provocation and can't be re-attached.  They're not cheap so it is a significant risk.

[TiTanecd]

Thanx for the explanation. 

How does a pure white image on a laptop (LCD as opposed to LED) compare. Have you tested LCD/LED to see the comparison from your other tests? Cleaned screen + white (using #FFFFFF)?

I would give this a go first before trying to then spend money. I didn't know it would be more technical, than a even white field behind a layer or 2 of panels.

[IanL]

Thanx for the explanation. 

How does a pure white image on a laptop (LCD as opposed to LED) compare. Have you tested LCD/LED to see the comparison from your other tests? Cleaned screen + white (using #FFFFFF)?

I would give this a go first before trying to then spend money. I didn't know it would be more technical, than a even white field behind a layer or 2 of panels.

I'm all in favour of trying cheap/free options before spending money!  I've never tried a laptop screen as I have no convenient way of pointing the scope at one during the imaging session; generally I wait 'til the next day to take the flats so using a simple diffuser works for me (but there is a risk of disturbing the imaging train or dust moving around, wrecking the flats).  Plenty of people report that they use computer screens with white images (or tablets) but I haven't seen anyone post a serious analysis of the light output.

Gut feel is that it might be okay since clearly a screen has to be able to reproduce a wide gamut of colours; it would be easy enough to test - just take a few flats and:

- Check the colour of the flats visually - if there is a strong colour cast then you may have a problem.

- Chek relative brightness and noise of the three channels to see if they are roughly the same or within an acceptable margin.

It's got to be worth a try if you can find a way to point the scope at the screen keeping the objective and screen parallel to avoid a gradient, and blocking out any external sources of light that might leak in round the sides.

[Bizibilder]

This is one of last night's flat frames taken using my new LED "flat light panel".  The camera is a modded Canon 1000D. I have included the histogram that shows three reasonably similar channels (cropped from Photoshop):

And with the colours aligned:

[IanL]

This is one of last night's flat frames taken using my new LED "flat light panel".  The camera is a modded Canon 1000D. I have included the histogram that shows three reasonably similar channels (cropped from Photoshop):

What LEDs did you use Roger?  At some point I might want to have another go at this project, so would be useful to know.

The fact that it is a modded camera will make a fair bit of difference.  The unmodded Canons (like mine) block a lot at the red-end which is another reason why I struggled with my own LED box in the red channel.

Regarding the colour alignment, as mentioned before it is not the lack of alignment that is the issue, it is the increased noise in the underexposed channels that causes the problem.  Yours look pretty close so hopefully not a big issue - mine were way worse than this (red was probably 1/3rd as bright as the other two channels).

[Bizibilder]

I use one of these:  http://www.ebay.co.uk/itm/Huion-A4-Ultra-Thin-LED-Animation-Drawing-Stencil-Board-Table-Pad-Light-Box-DE-/321458540668?pt=UK_Crafts_DrawingSupplies_EH&hash=item4ad86c107c

Overall considerable cheaper than buying all the bits and assembling my own!!

[D4N]

Anyone had any success using an iPad for flats?

TSED70Q, iOptron Smart EQ pro, ASI-120MM, Finepix S5 pro.

[TiTanecd]

If you take say 15 lights, do you align each channel for all 15 or leave as you took them? This is taking into account if the noise, is not too much. Would DSS remove some of the noise when creating the master?

The final result should (when stacked) remove vignetting (from the edges) and keep the centre (the part least effected by vignetting) the same as if you didn't have flats? I have also seen they remove dust motes, but I haven't seen or even know what they look like on my basic pictures.

[IanL]

If you take say 15 lights, do you align each channel for all 15 or leave as you took them? This is taking into account if the noise, is not too much. Would DSS remove some of the noise when creating the master?

The final result should (when stacked) remove vignetting (from the edges) and keep the centre (the part least effected by vignetting) the same as if you didn't have flats? I have also seen they remove dust motes, but I haven't seen or even know what they look like on my basic pictures.

Several questions there:

1. You should use the same light source and exposure length for all the flats, don't mix and match different sets of flats.

2. You need to expose the flats so that the pixels reach somewhere between 1/3rd and 2/3rds of the maximum brightness (i.e. the histogram peak should be 1/3rd to 2/3rds of the way across from the left side).  Different software seems to have different results and so you may need to experiment to find the best spot.  Using a DSLR you can set it to AV mode and it will usually come up with a usable exposure.  Otherwise you shoot in manual mode and change the exposure length whilst taking test shots until you get what you need.

3. Watch out for really short exposures though - a very fast shutter speed may end up creating dark bands on the flat so check for those.

4. Personally I wouldn't bother trying to equalise the colour channels in the flat frames.  If you don't you may (or may not) end up with a colour cast in your calibrated image, but you can fix any colour imbalance as part of the processing (even with perfect flats you will probably still have to tweak the balance anyway).

5. You can't remove noise from an image - noise is random by definition.  When you stack flats (in DSS or whatever) you increase the Signal to Noise Ratio (SNR) - both the noise and the signal (the flat background) increase as you stack more frames, but the signal increases faster than the noise as you add more frames.  It might be seem a subtle point, but it is worth knowing.  Basically the more flats you have the better - I'd go for at least 30, but if you have time and the software can handle it then go for more than this - after 100 or so you will see little real improvement so don't go mad.

6. Flats will remove the vignette caused by uneven illumination, but for this to work you need to make sure the camera, focuser and any filters that you might use (e.g. LP filters) do not move between the lights and the flats.  You'll often find the vignette is not centred in the image, so any rotation will end up creating odd coloured fringes in the image.  Any re-focus will do similar as the size/brighness of the vignette will change.

7. Flats should also remove dust shadows from dirt on or just in front of the sensor (bear in mind anything on the mirror or objective will be so far out of focus as to not be visible).  These are either little round dark spots, dark "doughnut" shadows or sometimes fat dark wiggly lines caused by fibres.  Again the trick is to ensure the dirt doesn't move between darks and lights.  If your camera has an auto-sensor cleaning routine (most modern cameras like Canons do) then turn it off when doing astro-imaging.  Here's a nice image I took which was wrecked because I forgot to do so.  The dark circles (e.g. bottom right) are the dust shadows on the light frames, and the light circles are different shadows on the flats where the dust has moved - the flats are over-correcting the image at these points and so creating 'inverse' shadows:

(Click image to enlarge).

[Bizibilder]

Don't bother aligning the channels - DSS will do that for you - just bung them in and let the software do its thing.  Dust motes (or dust bunnies) look like smudges or fingerprints on the image - you'll know one when you see one!  They generally come by the dozen  !!

[TiTanecd]

Thanx guys, it helps alot. 

I'll try get some flats next week when the weather clears. And for learning I'll run 2 tests, 1 with flats and 1 without to also fully grasp what is going on. 

[TiTanecd]

I took 30 Flats with my LCD, this is a single shot with a histogram. To me it doesn't look like the one above, its a lot wider.

This is the Master Flat. Does this look like it should?

This is stacked with all recommended settings and all frames (Bias and Darks). It Has the opposite effect than I'm looking for. Is this due to the fact that the Flats aren't pure white?

From what it looks like this is wrong. Mine go very black in the corners where the one above has a slight darkening.

These were only 1 min exposures at 800.