You're right Lee, I did mean sky limited not RN limited. I had the same idea in my head, just the wrong wording
Looking up the sensor that goes into this camera, used in a bunch of Olympus models, it appears to have an peak QE of 57%. This puts it on par with most of the front illuminated KAF sensors.
Only thing about that absolute QE is that it's a OSC version of the sensor, which will obviously impact the QE. I'm not sure how consistent the difference in QE between mono and bayered variants is, maybe it's sufficiently consistent to get a good idea, but without further information I wouldn't like to guess.
You're right Lee, I did mean sky limited not RN limited. I had the same idea in my head, just the wrong wording
Looking up the sensor that goes into this camera, used in a bunch of Olympus models, it appears to have an peak QE of 57%. This puts it on par with most of the front illuminated KAF sensors.
Were those mono sensors? As far as i can determine Olympus did not make a dslr with the mono version of this Panasonic sensor. Sensorgen has figures, as do other sources like DPReview but they are all for the colour sensor. The Relative QE charts i have seen do place the 8300 and 1600 very close but the 1600 is ahead on lower noise and slight better QE in the middle of the spectrum. Still even if they were equal on all aspects of performance, the 1600 is over $1000 cheaper than the lowest cost 8300 camera. I have to wonder how long Atik and others with 8300 cameras can maintain a big ptice premium.
I don't believe that the absolute QE really changes between OSC and mono but it is correct that these are the first mono series of this particular sensor.
Comparing the 1600 against its closest rival, the 8300, the 1600 at this stage appears to be better in virtually every respect. It has shallower wells but much lower read noise rendering it with a much higher dynamic range. As you point out Glen, it is also FAR cheaper than any 8300 on the market.
The 8300 is more expensive because the CCD technology is more expensive to manufacture but I am with you, I don't know how much longer this particular one will last into the future. Probably until the 1600 proves itself to be a reliable replacement
What I was getting at with comparing the new 1600 against a 16803 is that it isn't a replacement of noisy KAF sensors
I go away for a few days and there's an extra 4 pages in the thread! Time to get comfy and start reading!
In all seriousness though I am thoroughly enjoying reading this post. Really appreciate the effort Glen, Ray and others are going to with sharing their first experiences with the ZWO and other relevant cameras.
Thanks guys! .
To prevent highlights from blowing out in a long exposure perhaps?
From what little I have seen of this camera it seems to be its Achilles heel.
Isn't blowing out the highlights always going to be a problem with chips with increasing sensitivity? I'd never thought of sensitivity as being a weakness before
Quote:
Originally Posted by Atmos
As you increase the gain it does drop the read noise but the overall dynamic range drops with it. i.e. the read noise falls slower than the decrease in well depth.
Yes you can image at gain 300 with 1s exposures in Lum (gregbradley noted that it clipped the core of M51 even at 1s). Using a 3nm Ha filter, if you can drop the read noise to 0.6e- you need ~50s exposures (at F/5) to be RN limited but your well depth is so shallow that even at that exposure time you're clipping.
To give a bit of a comparison, at gain 0 (where the 1600 has its highest dynamic range) compared to a KAF-16803 on the same telescope, they need about the same exposure time to become RN limited. This is given that the KAF has a RN of 8e- (I believe that RickS mentioned a while back that this is what his PL16803 gets at SRO), this is within a margin of 10% (the 16803 reaches RN limited a fraction quicker without taking dark current into consideration but the 1600 is likely to be lower in that regard) . The 16803 has 5x the well depth however so while the 1600 will clip stars and bright cores (thinking of lagoon, M31, M42 ect) the 16803 will barely notice the difference.
So is the clipping stars a problem a result of the dynamic range/well depth or the sensitivity?
It seems that we all want the perfect chip but this isn't it, but then nothing ever will be. The 1600 sounds pretty impressive, my perspective being to use it as an alternative to an uncooled DSLR and more expensive astro CCD.
Have we figured out a way to get more than 12-bit data from the camera?
My angle is how we may need to adapt the way we image to the CMOS generation.
Have we figured out a way to get more than 12-bit data from the camera?
My angle is how we may need to adapt the way we image to the CMOS generation.
get as much bit depth as you want by stacking:
1 sub = 12 bits
2 subs = 13 bits
4 subs = 14 bits
8 subs = 15 bits
16 subs = 16 bits
etc...
since you can do hundreds of short subs, the sky is the limit. 12 bits is not an issue at all
just stacked 19x 30 second subs and 10x 60 second subs of m83 to see how much of an SNR hit there was going to the shorter subs. I was using gain 100 and calculated that 1 minute should be about right. Images below show a linear mapping of the bottom 1000 ADU for the 30sec and 2000 ADU for the 60 seconds - no attempt was made to non-linear stretch, so everything bright is saturated.
Looks like the calculations were about right - the 60 second data looks good, but the 30 second data is down a bit on detail (seeing was worse as well, but contrast is down a bit). I had thought that 20 second subs would be OK and it still looks that way, provided that the gain is pushed up to maybe 200 to get the read noise right down.
By the way, there was no calibration or processing of these images - the camera is as smooth as.
I have just finished an unattended narrowband run on M16, SGP ran it all, including managing the cooling of the sensor to -25C at the start and warming it back to ambient at the finish. I checked on it once in a while, half expecting it to stop but no issues at all.
Captured 25 x 60" Ha, 25 x 60" Oiii, & 25 x 60" Sii, all shot at Unity setting (Gain 139, offset 21). Watched a few of the files come up on the display and the histogram did not seem to show any clipping, but i probably need to process the stacks to get a idea about that. I will stack and colour process tomorrow morning, should have the image up on Astrobin later. Beautiful clear dark night here, wish i had some RGB filters as the Helix is now in a great position, soon i will get to that.
Finished a quick process on the M16 narrowband from last night. A number of problems (probably mostly due to me) with the resulting image. First, despite 25 x 60" subs of SII there just doesn't seem to be much of it there at all, could be this target but maybe it just needs more time. I can see it in the stacked SII image but bringing it out is almost impossible without destroying the image; so I have left the SII out for now. What I have posted to Astrobin is a Bi-Colour version (with Ha - Red, and OIII - Green). I tried a Hubble pallet and was not happy with it and it needs more work.
I know there is more there but am having trouble bring it out without making a mess of it.
These were shot at Unity (Gain 139, Offset 31 (I think or was it 21)). Looking at it now on Astrobin I am thinking the stars look too green and I need to turn the saturation down a hair but not now. The Astrobin histogram looks pretty pathetic. Generally I think it is too dark as well.
I will stick a small one on the bottom here but have a look at the big one and give me your opinions on how I could do it better:
Should be enough SII there at 25 minutes. As a comparison, I have an even shorter integration, single 900s SHO (45 min total). http://www.astrobin.com/full/245969/B/
Yeah Colin, but you know what your doing! I will have another go at the Sii. It would be nice to see what you could do with these files. Maybe i should outsource my processing.
Glen, the CN guys have been talking about this camera having a lower Red response than expected. If the QE is lower at the red end of the visible spectrum it could explain the poor response in SII
If your SII actually looks okay without being combined then it is just a question of colour calibration. The SII when stretched can look pretty good but it is quite weak compared to the OIII and especially Ha. For instance, you may have an area with a pixel value of 3200 in red, 18000 green and 11000 in blue. It will appear a greenish colour, this is why a SHO when just combined looks largely green. The Ha signal is overwhelming.
The idea is that the SII (red) needs to be stretched far more to get the correct Hubble Palette. This is the reason for magenta stars, the red and blue get stretched more than the green for colour balance.
Richard, in regard to the spectrum response, Ha is down there are well and I have plenty of Ha, Sii is not much further along. However, when I look at the spectrum chart on the ZWO website I can see that it is less responsive in that part. https://astronomy-imaging-camera.com...si1600mm-cool/
But since Ha is so good, unless the cover glass is a cut filter that is affecting Sii spectrum, then I am sure it is just something stupid I have done along the way. I have just been playing around stretching a single Sii sub and it has more signal than my stack of Sii appeared to have. Even on the SGP display I could see signal on individual Sii frames. I need to go back through what I did in DSS with that stack.
Uodate: looking back at my Trifud SII data it was pretty skinny as well when shot on equivalent integration times to Ha and Oiii. I will test tonight on M16 again running just Sii to build up more signal. Is there a 'standard ratio'?
FYI, the attached table shows predicted optimum settings for my 250f4 system for a 3hour broadband galaxy imaging session under fairly dark sky. Early testing shows that this may be slightly optimistic - 1 minute subs gave better results than 30 seconds at gain 100 in the real world.
The most interesting outcome is that the dynamic range holds up well up to gain 100.
Yeah Colin, but you know what your doing! I will have another go at the Sii. It would be nice to see what you could do with these files. Maybe i should outsource my processing.
hi glen - I'd be keen to have a crack - mostly to check out the 1600 files and because the weather has been so bleak I'm getting rusty.
Thanks Russell, the three narrowband files are over 30meg each before i register them against each other. My Sii issue is probably just a lack of signal and i will attrack that tonight. I will see about putting a file in the file sharing box at Astrobin, so you can have a go.
FYI, the attached table shows predicted optimum settings for my 250f4 system for a 3hour broadband galaxy imaging session under fairly dark sky. Early testing shows that this may be slightly optimistic - 1 minute subs gave better results than 30 seconds at gain 100 in the real world.
The most interesting outcome is that the dynamic range holds up well up to gain 100.
Ray perhaps i should drop my gain setting from unity ( g= 139) to g=100. I am shooting 60" subs on my test targets (nebs and such).
Thanks Russell, the three narrowband files are over 30meg each before i register them against each other. My Sii issue is probably just a lack of signal and i will attrack that tonight. I will see about putting a file in the file sharing box at Astrobin, so you can have a go.
cheers mate, yep s2 is always a pain.
just for info, I am considering the 1600 to match it with my goto dob (12" f5). the plan would be to still use the EQ8 (12" f4) and 694. but use the eq8 for Luminance (694) and colour gathering (1600) on the dob.
I am unsure on how narrowband would work on the 1600 with 20-25 second subs. so it may only be a rgb solution.
Ray perhaps i should drop my gain setting from unity ( g= 139) to g=100. I am shooting 60" subs on my test targets (nebs and such).
if you are shooting narrowband, maybe try much longer subs Glen. I haven't tried, but was thinking of using 5-10 minutes at about gain100, although 139 should be pretty much as good.