Flats issues

[bottletopburly]

For my flats i have been using a EL panel behind frosted perspex all seemed fine until lately and after processing seemed to have lines running across screen a bit of playing around  i found the issue was the flats and more so the EL panel ,now i dont know if the EL panel has a fault or the power supply causing the issue maybe some sort of refresh rate issue ,has anyone had issues with using EL panels , i use a Nikon D5300 and image in APT ,now using AV mode using the 5300 is not as simple as a canon  when no lens attached so measure exposure in AV mode in APT then manually set a flat plan a bit long winded but simple once you know the exposure for your light source ,i would like to measure the ADU of my flats can anyone tell me how if  can be done using Siril ?attatched files show I/40 sec using an EL panel showing red as i have used a sheet of pink paper and two sheets of tracing between El panel and perspex , 1/500 sec from a tablet screen and a previous shot of m52 showing the lines i proceess in siril  and Dither so i only use lights bias and flats when imaging ,So could someone take a look at my flats and tell me if they are any good is the EL panel flat any good compared to the tablet Flat .

 

Single__0001_ISO200_1-40s__NA.NEF Single__0002_ISO200_1-500s__NA.NEF

[david_taurus83]

Nothing simple about Canon flats either as Av mode doesnt work at all for me. Always under exposes, even if I set the ev to +2. I always aim for flats to be at least 1 second and aim for around 2s. I then check in Pixinsight for the mean ADU value to be around 50% of the cameras max ADU. My 14 bit 6D is around 16k so I aim for 8/9k mean value.

[davew]

The flat taken with the EL panel looks bad. Mainly for the banding which doesn't appear on the other one. Another thing bad about it is the fact you've done it through a red sheet. You may already know this but using that coloured sheet is not a good idea as the signal to the red pixels is way out compared to the green and blue. That means no matter what your light source the flat won't work properly.

On the few occasions I've done flats with a DSLR, I've used a evenly lit white wall. I only used a light panel with a cooled CCD camera. 

Dave.

[bottletopburly]

Cheers @davew I only put the red on whilst trying to see if putting sheets in front of panel would help I’m wondering if it’s being caused by the power supply or the actual panel  , I can use the tablet for refractor for the 200pds I can point at the living room window until I sort panel issue out .

[davew]

Ah, I see.

Regarding the panel ... I haven't got a clue what the cause of the banding is but I wish you the best of luck finding out. Does the banding still appear if you block most of the light with sheeting and expose for much longer ? Does the banding appear if you use a lens closed right down and also using a longer exposure ? Is the camera set to 60 Hz by mistake ?

Questions questions ! 🤔

Dave.

[bottletopburly]

1 hour ago, davew said:

60 Hz by mistake ?

actually it is will this be an issue

[davew]

Possibly. GB runs at 50 Hz. Change it over and have a test but it may make no difference.

Dave.

Edit. This normally affects video but you never know

Edited April 4, 2020 by davew

[bottletopburly]

@davew  @david_taurus83 

So before i bin the EL panel i have now put two A4 sheets between perspex and EL panel and followed the advice in article https://www.myastroscience.com/proper-flats-with-dslr so a 20 sec exposure gives a mean value of 16383.0  so  a bit of experimenting with exposures i need half that and a 1.2 sec exposure gives a Mean value of 8055.5  can anyone tell me is the 1.2 sec looking any good for a Flat ,using AV mode does not work properly.

 https://www.dropbox.com/s/znfszypsbnuwart/Single__0003_ISO200_20s__NA.NEF?dl=0

https://www.dropbox.com/s/l9n7kybunsq1h1g/Single__0007_ISO200_1.2s__NA.NEF?dl=0

[michael8554]

Here's how the 1.2 sec image looks in PSP, the histogram is showing the Lum position :

So way past the midpoint, and blue almost saturated (246 out of 255)

So I'm not believing the statement in the link that the Canon AV setting is wrong.

Michael

 

 

Edited April 7, 2020 by michael8554

[bottletopburly]

Thanks  @michael8554 so not an accurate article then , I will have a tinker .

[michael8554]

This business of midpoint exposure is tangled. 

On a Canon, the AV exposure viewed on the lcd screen is midpoint in the histogram, but that view has been  processed and stretched from the Raw into a sort of jpeg, for viewing purposes. 

The Raw will look very dim if correctly viewed. 

So depending on your image editor or PC viewer software you may get varying results. 

So include me in that caveat, my image may not be accurate either! 

I'm not sure if I've already said, try an exposure that is midway on your Nikon's histogram. 

Michael 

[bottletopburly]

27 minutes ago, michael8554 said:

try an exposure that is midway on your Nikon's histogram. 

Will do Michael thanks 

[david_taurus83]

Your 1.2s flat looks good to me.

 

For comparison, here is a couple of Av mode flats from my 6D with a 50mm lens at f1.8 pointed at a white ceiling. This first on is Av at 0 Ev.

 

A mean value of 2723 ADU out of a possible 16K ADU. Now, if I calibrate that flat with a master bias file which on a Canon is 2047 ADU then that doesnt leave much of a flat, does it??? It simply doesnt work!

 

Here is another Av flat but this time the Ev is set to +3.

 

A much better value and even when the bias is subtracted I'm still left with a value which is approx 1/3rd of the available ADU.

 

Ultimately, I suppose, its whatever works for you. For me, Av flats under correct and emphasize dust bunnies in the final stack whereas the method I use (similar to the article) removes them.

 

[michael8554]

Interesting how PI sees the 1.2 sec image.

The Screen Transfer Function window shows L at midpoint. So in PI the 1.2 sec image appears to be correct.

The RGB values are also way different to my display of the image

But at least the main window showing the Raw is very dim and doesn't look to be at midpoint, as I explained.

Now on my 6D I dither and use the Bias as Dark, and the AV flats work very well and removes bunnies.

But there's no point me arguing on which method is best, it's what works for you Bottletopburley that matters.

Michael

 

[david_taurus83]

Well I have to hold my hands up to this and eat humble pie. I took some flats a couple of nights ago and ran NINA's flats sequence. It gave me a very low ADU result so I cancelled after 5 flats and then done a manual set with my target ADU around 8/9k as described above. I tried stacking the 3 hours of data I have of M51 in DSS and my manual flats did nothing apart from enhance the vignetting and dust bunnies. So I created a master flat out of the 5 NINA flats and they work? Guess you cant trust the numbers all the time?! I've changed scopes for M51 and not using a flattener on the 102 f7. So maybe the numbers the flats need to work aren't the same as on the 70mm quad.

[michael8554]

Hi David,

That's interesting, what ADU were the ones that worked ?

Michael

[bottletopburly]

@david_taurus83 was that ccd or  dslr? 

[david_taurus83]

Canon 6D. The ADU value if I use the 70mm quad f4.9 and NINA to shoot flats is around 8k usually. With the 102mm F7 NINA flats are around 2.5k ADU. The odd thing is the CR2 thumbnails look virtually the same if opened in windows picture Viewer. The difference between both scopes is the quad produces no vignetting at all as it does correct for a full frame chip whereas I'm using the 102mm with no flattener so there is pretty severe vignetting so I'm wondering if the light fall off in the corners has an affect on the overall mean value. Guess I'll just have to trust NINA to produce successful flats and I'm going the 5 it produced on my first night on M51 ste enough to salvage 3.5 hours of data!

[michael8554]

19 minutes ago, david_taurus83 said:

The odd thing is the CR2 thumbnails look virtually the same if opened in windows picture Viewer.

Yes, the psuedo JPGs that camera LCD's and PC's display have no real relationship to the actual Raw.

Out of interest could you post one of your 2.5K Flats please ?

Michael

[bottletopburly]

I tried Nina flats wizard if I got it right inputting iso 200 and leaving everything at default setting it worked out 0.5 sec  a longer exposure than I would normally use , I have a replacement flat panel coming this week hopefully so will wait and see how that measures .

[david_taurus83]

On 14/04/2020 at 11:21, michael8554 said:

Yes, the psuedo JPGs that camera LCD's and PC's display have no real relationship to the actual Raw.

Out of interest could you post one of your 2.5K Flats please ?

Michael

I've yet to process the 102mm data but I processed the last project I done on the 70mm quad. I experimented with various flats ranging from 2k to 9k ADU. In all circumstances the master flats cleared the individual subs of pretty bad dust bunnies on inspection. However, after stacking 180 subs there were still evidence of the worst dust bunnies on the data calibrated with 4k flats. So I redone everything again with 9k flats and they cleared everything.

With the 102mm though, I have a feeling I messed up the settings in NINA. I was using my laptop, not my usual imaging mini PC and looking at the settings used I have 16bit for the camera profile. So unsure if NINA scaled the ADU value for those flats. Might end up dumping 3.5 hours of data!!

[bottletopburly]

@david_taurus83 i got a tracer panel though not impressed with quality of it they illuminate from one edge so this week i will make my own light box , but i do like the nina flat wizard though

[david_taurus83]

3 minutes ago, bottletopburly said:

@david_taurus83 i got a tracer panel though not impressed with quality of it they illuminate from one edge so this week i will make my own light box , but i do like the nina flat wizard though

Yeah I bought one of those when i started. Wasnt convinced either so I ended up splurging on an Artesky panel. Expensive but worth it! That said, I had forgotten how hard it was to get good flats with a DSLR since I had a 600D. When I moved to an ASI1600 mono it was soo easy to do darks, flat darks and flats.

[bottletopburly]

@david_taurus83Test flat using nina on the newt ,when i stretch it in Apt you can see the newt rings if it shows the rings off centre to the image centre  can anyone tell me  does that mean your collimation is out ? https://drive.google.com/file/d/1Iqee9i0fFxQNkn7GlRv12ytrwKWUAbTE/view?usp=sharing

Edited May 7, 2020 by bottletopburly

[PadrePeace]

I too have been wondering about the exposure I'm using for my Flats on an ATIK 460EX mono. Until recently I had been letting the Flats Generator in APT do the leg work with a dim EL panel with a frosted 4mm clear diffuser sheet as the light source. That typically gave me 0.01467s for a Lum 1x1 Flat looking to hit 20,000 ADU which is the Atik recommend full well number. 

I've been told that may be far too short and that some recommend >1sec exposures for Flats. To achieve that I had to put six sheets of good quality A4 white paper under the EL panel. It was so dim I could hardly see the light myself. Whilst that worked for a 1sec Lum Flat my OIII time shot up to 87s which I know is because its NB and 3nm at that but that becomes a 45min exercise to catch 30 Flats.  

Using a technique suggested on the APP sight I had the histogram results posted below:

The first is for a single 1x1 Lum Flat at 0.01467s exp
 
The second for a single 1x1 Lum Flat at 0.205s exp

I went with 0.2s compromise ove the full 1sec as that brings the OIII down to just 12 sec exposure.  Considering the negligible variation and that both are nicely positioned at the 1/3 point with no clipping why should I need to doubt my existing settings (0.01467)? 

Finally, what would you guys recommend for this specific CCD camera as a target Flats exp setting and why?

kind regards

Chris Peace

PS: Just a thought not requiring an answer unless anyone feels so inclined but if I need to get my light source so dim, why are bespoke Astro EPs and light boxes, sold expressly for making Flats, so bright in the first place?

Edited May 14, 2020 by PadrePeace