ATIK 383L+? Very dissatisfied!

[gorann]

20 hours ago, Magnum said:

Id agree with most of that, but we can't forgo dithering with cmos, as they all have a faint fixed pattern noise that CCD doesnt have. you can see this become evident if you stack a load of bias or darks from even the latest zero amp glow models, a pattern quickly emerges. In real world imaging with the 533 I too dont use darks or bias, but if I dont dither I get diagonal banding pattern to the noise, ie walking noise which I dont see with my CCD's. luckily dithering completely solves  it and I always dither with either CCD or CMOS, so yes the 533, 2600 & 6200 are probably the closest cmos has got to CCD so far.

Lee

Walking noise was always a problem with my DSLRs (Canon 60D and 60Da) but I have not noticed it with my two 2600MC, and not even with my older ASI071, and I try to go as deep as possible in many of my images, especially when shooting IFN. Maybe it varies a bit between individual cameras?

@AbsolutelyN noticed a very wide banding noise in his ASI2600MC and it was discussed in a thread here. It was very wide so nothing that could be cured by dithering. I sent him a flat from one of my 2600 and he found a similar but weaker noise after stretching and running it through DBE. At the end of the thread @wimvb suggested that it was Newton rings possibly created by the optical window or the glass on the chip, so not of electronic origin.

 

[whitfieldp]

My 6 year old QHY9 mono has the same KAF8300 sensor that I run at -15degC. I tested -20degC and there was minimal added reduction in noise for me at least.

My darks have no-where near that number of hot pixels, even with 600s at higher gain for narrow-band or binned 2x2. I don't have a stretched jpg to hand and my APP masters are 32-bit but I could probaby rustle something up if anyone is really that interested. At room temperature I vaguely remember a column or two but they disappear on cooling. Maybe I got lucky but it doesn't seem to be inherent to the sensor chip.

I'll keep using my QHY9 until it dies. The final images look good and the pixel size is a perfect match to my optics. Alot of CMOS sensors would be way oversampled and the 16-bit ones are still a bit pricey at the moment.

Pam

[Magnum]

On 21/04/2021 at 00:39, Magnum said:

Thought I might as well sow the quality of data the 383 produces when used at 13.8volts @ -20C, ive attached screenshots of uncalibrated single 900 sec subs of the Pelican and Heart nebula's both at 100% to show how clean the subs are, and also the final images which are both only 15 x 900 secs each using just bias frames and dithering.

Please don't give up on the camera unless you want to give it to me really cheap LO 😛

Lee

 

 

Replying to myself here as I forgot to mention that most of what look like hot pixels in the 383 darks are not true hot pixels at all, if I measure mine 95% of them read around only 900-2000 ADU So that would be considered a warm pixel, a true hot pixel would be maxed out at 65,000 ADU.  Now If you take a Dark it can look scary seeing lots of bright dots against black, but as you can see in my single 900 sec sub above on a real target,  most of them aren't visible at all and the ones that are visible are clearly not white and dimmer then most of the faint stars. 

This explains why they soon stack out with no special intervention on my part, just a very slight dither and stack with median or SDmask in maxim. 

Conversely when using the Sony sensor in my Atik 460, yes it only has maybe a dozen hot pixels, but they tend to be true hot pixels ( ie maxed out 65,000ADU ) that take much longer to stack out and require more dithering to do so or more subs in the stack.

I think this is why a lot of the experienced ccd imagers Like Olly and Sarah Wager really appreciate the Kodak sensors. I can can literally make a perfectly clean image from 4-5 long subs with the Kodak sensor in my 383. ignore what the darks and focus frames look like just trust the data and you will be amazed at its quality. 

As much as I like my cmos ASI533MC I could never make a final image with it in only 5 subs lol, it needs at least 50 -100subs to get really smooth. Problem with that is it uses a lot more storage but even worse it takes many times longer to stack, and as im probably the laziest imager ever I find that rather tedious. Having said that, the 533 is probably a better camera than any OSC ccd ive used for trying to produce a colour image in 1 night, but not a match for the Kodak mono chips in ultimate quality of the data.

sorry im starting to turn this post into a cmos vs ccd debate, I will just say on that they are both good technologies, I just get sick of the FB crowd from a certain uk retailer saying CCD is dead 😛 well if they dont want them I will gladly take them all off their hands 🙂.

Lee

 

[gorann]

The CCD cameras are of course not dead, it is just the market for them that died, so Lee, you should go shopping and give yourself a Christmas present, maybe get more real estate than the 383🧑‍🎄

[Magnum]

1 hour ago, gorann said:

The CCD cameras are of course not dead, it is just the market for them that died, so Lee, you should go shopping and give yourself a Christmas present, maybe get more real estate than the 383🧑‍🎄

The amusing thing is, although the people that say CCD is dead also say that they are worth hardly anything on the used market, yet in fact prices of used CCD's seem very stable and if anything have actually gone up recently. I know I cant replace my 460 and 383 for less than I paid for them 2 years ago. Despite those people trying to tell us otherwise in a vain effort to try and make it happen.

Same people also say that the large pixel size of CCD is unsuitable and grossly under sampled when using short focal lengths like camera lenses  which will result in awful blocky stars 😛  well here are 3 taken with 135mm lens at 9"/pixel , looks ok to me even though those people tell me im wrong 😛 

Sorry ive hijacked the thread, but its still about the 383 LOL

Edited November 4, 2021 by Magnum

[ollypenrice]

Out of interest, I do have an issue with my 460. I struggle to get the background sky up to the 23 I like to see, as measured in Photoshop. What I find is that my background sky is made up of pixel values ranging from a lot less than about 21, up to and including 21. If you like, it's more a cool pixel problem than a hot. My workaround is to get the brighter background pixels up to 21 and then pin the curve at 21, put fixing points above that and stretch below it, not hard enough to remove all grain.

I image at a dark site, which might be relevant, and still like my Kodak chipped camera since saying I did early on this thread. I'm not using it because I've lent it to one of my robotic clients but it's still beavering away.

Olly

[Magnum]

3 minutes ago, ollypenrice said:

Out of interest, I do have an issue with my 460. I struggle to get the background sky up to the 23 I like to see, as measured in Photoshop. What I find is that my background sky is made up of pixel values ranging from a lot less than about 21, up to and including 21. If you like, it's more a cool pixel problem than a hot. My workaround is to get the brighter background pixels up to 21 and then pin the curve at 21, put fixing points above that and stretch below it, not hard enough to remove all grain.

I image at a dark site, which might be relevant, and still like my Kodak chipped camera since saying I did early on this thread. I'm not using it because I've lent it to one of my robotic clients but it's still beavering away.

Olly

That's interesting Olly, when you say background level of 21 do you mean ADU or photoshop levels? I presume the later.

I always pre stretch my images in MaximDl with DDP and make sure I have a big gap on the left of the screen stretch, then save to tiff before opening in photoshop so the background is already at a high starting point then I bring it back down at the end of processing.

Be interesting to see what the average ADU of your bias frames are, mine are about 350ADU with my 460, if that helps?

Lee