290 replies to this topic

[MvZ]

Hi All,

 

Just a quick report of my first impression of an ASI224 camera with IMX224 cmos by Sony. Unfortunately, I have had pretty bad weather last weekend, and the few deepsky images that I did make still need to be processed, so for now this is just a post with some technical details.

 

The camera is a color camera, unfortunately, there is no monochrome version of the CMOS available. The camera has 1280x960 pixels that are 3.75um, and an USB3 interface. It can do about 150FPS in full res, faster in ROI. IR sensitivity is better than ASI120/174.

 

I measured a read noise of between 1.5 and 0.75 electrons (Depending on the gain used).

This is extremely low! (ASI174 and ASI120 has between 6.5 - 3.5e, ICX618 between 14 - 7e). This makes the camera very interesting for short exposure deepsky, or imaging very faint targets without having to fall back to longer recordings (and less 'sharp' frames), or for example Venus in 1000nm IR). CH4 imaging at 889nm should be possible without bayer grid artifacts, so you can use the camera as a monochrome camera then, as all pixels show very similar response above +- 810 nm.

 

I only made a few images over the weekend, as the weather was poor and there was a LOT of wind (a big Dob out in the open does not help here). I did one test run of M57 with 0.1 second exposures, because anything longer resulted in blurry frames due to the wind. I also did some 1 second exposures of M51 to try and get some color into my M51 image I made a week earlier. I still need to process the results, but for the M57 I need to change AutoStakkert!2 as aligning thousands of these frames proved very tricky. But I was really impressed with the sensitivity of the camera so far, and got quite a lot of detail from the M57 image (not deep, but very promising for just 0.1 second exposures).

 

I hope to do some more actual imaging with this camera in the coming months. I'm sure Sam will come with more details regarding this camera soon, as well.

[yock1960]

Sounds interesting! Looking forward to seeing the images!

 

Steve

[John Boudreau]

Figures something like this would come along right after I get an ASI174MM.

 

Seriously, it sounds like a great all-around performer. As for the possibility of it for CH4 imaging of Jupiter or Saturn, I'm sure it would work--- I own an ASI120MC that gives cleaner CH4 images than my ASI120MM largely due to the near-IR Bayer filter leaks.

I haven't found any QE charts yet for this sensor. I'd be curious to find out if it can do UV imaging of Venus as typical Bayer filter transmission is poor in that region. Time will tell, but it may still be a better answer for our pursuits if it becomes available in monochrome.

[MvZ]

> I'd be curious to find out if it can do UV imaging of Venus as typical Bayer filter transmission is poor in that region

 

Good point, I'll test that next time!

I did see a (relative) QE chart, but it is difficult to interpret the results, as it only shows from 400 to 1000nm. I should just do some actual recording on Venus, and compare it to the ASI120MM for example. Not sure when I have time though (and it would help if the weather cooperated too)

 

I doubt a monochrome version of this camrea will come out. AFAIK, there is only a color IMX224 available.

[Peter in Reno]

Are you Beta testing this camera since I cannot find it at ZWO web site?

 

Are you seeing any Fixed Pattern Noise (FPN)? Does the camera come with USB 3.0?

 

Peter

[MvZ]

> Are you Beta testing this camera since I cannot find it at ZWO web site?

Yes.

 

> Are you seeing any Fixed Pattern Noise (FPN)?

Very little, I will do some more testing later this week.

 

>Does the camera come with USB 3.0?

Yes. It can do 150 FPS at full res.

[Ethan Chappel]

Do you know approximately when this camera will be released?

[StrangeDejavu]

Following. Very interested in more information and some pics.

[Vars191]

I am VERY VERY VERY interested, where and when can I get one it is perfect for what I want to do

HALLEULLA  

[Peter in Reno]

According to this web site, planned mass production of this new CMOS sensor is December 2015:

 

http://image-sensors...ity-sensor.html

 

That probably means cameras won't be ready till sometime in 2016?

 

Peter

[MvZ]

>Do you know approximately when this camera will be released?

I'm not sure really. I'll ask Sam.

 

Regarding the fixed pattern noise, what is the best way to quantify this? I mean, I can see one vertical line at high gain, but other than that, the picture looks quite clear. I can measure the standard devitiation of stacked bias frames (or maybe better: the stdev of many bias frames averaged, and subtract the read noise component). Assuming that fixed patter noise is what is what is left over in the images when there is no signal and the read noise has gone? But there must be good ways to look at columns and rows too??

[wenjha]

ASI224 camera is the next generation camera!

its far better than all the camera available on market

here is a fast test compare with ASI224MC and ASI120MM

its not fair to compare between a color sensor and mono sensor

but the fact is this color sensor is even better than the mono one

 

the gain range of ASI224MC is 0-60db and ASI120MM is 1x-64x(36db)

 

attached the test results under CH4 filter and Clear filter

 

Attached Thumbnails

• 
• 
• 
• 

[wenjha]

>Do you know approximately when this camera will be released?

I'm not sure really. I'll ask Sam.

 

Regarding the fixed pattern noise, what is the best way to quantify this? I mean, I can see one vertical line at high gain, but other than that, the picture looks quite clear. I can measure the standard devitiation of stacked bias frames (or maybe better: the stdev of many bias frames averaged, and subtract the read noise component). Assuming that fixed patter noise is what is what is left over in the images when there is no signal and the read noise has gone? But there must be good ways to look at columns and rows too??

I think the easiest way, just need to attach a stretched bias frame

[MvZ]

Here is an inverted image of M57 I made by stacking 3300 x 0.1 second subs. You can see many faint stars, and even faintly see part of the galaxy IC1296 in the upper right corner.

 

Read noise this low is a game changer for very low light imaging (deepsky, OR faint planetary targets - or moons of planets)

 

Like I said, the conditions were quite terrible (lots of wind, and poor seeing), but with exposures this short it doesn't matter much.

 

Attached Thumbnails

• 

Edited by MvZ, 16 June 2015 - 06:38 AM.

[MvZ]

You can imagine stacking frames like the attached one (I debayered it for display purposes, but of course processed them raw to get the best possible resolution) is a complete PITA. There is pretty much nothing to track on, and I had to tweak AutoStakkert!2 a bit to make sure everything went 'OK'.

 

You can see that even though there is very little signal to work with, the low read noise allows to stack (many) frames like these and still end up with quite high signal/noise ratio.

Attached Thumbnails

• 

[TorstenEdelmann]

attached the test results under CH4 filter and Clear filter

 

Hi Sam,

 

20 vs. 2fps in CH4 ?!?

Was this test done using same image scale and scope ?

 

Torsten

[MvZ]

And the color result of M57 with a total integration time of 5.5 minutes (with the 0.1 second subs).

Attached Thumbnails

• 

[charotarguy]

And the color result of M57 with a total integration time of 5.5 minutes (with the 0.1 second subs).

Hi Emil,

 

That's is an amazing image, what scope was used to take this image, of course this has nothing to do on your expertise to take such amazing images. Can you also share the process of taking such images. It would be great to know. Thanks.

[schmeah]

Amazing! This could be a game changer. For both planetary and deep sky.

 

Derek

[MvZ]

The 16" Dobson was used, of course

 

>Can you also share the process of taking such images

Well, first I have to find the target, and typically I do this with Stellarium on my laptop followed by some starhopping (which is fun, but if you just stared on your laptop screen, it can be tricky). I use a 32mm 2" eye piece for this. M57 is easy to find of course, but some other deepsky targets can be much more tricky.

 

Then I switch the eye piece with the camera, move the focuser inwards - two turns or something like that - and I'm relatively close to focus (e.g., I typically then am able to see blobs of stars with 2 second exposures and high gain settings). I fine-tune the focus by switching to extremely short exposure times (typically I get 10fps) and paying close attention to one single star that is not over-exposed (so you can really see it 'shine' when it is perfectly in focus).

 

Then I put the camera settings correct (e.g. relatively high gain, but with this camera certainly not max. Just to have pretty much the best read noise performance), and start recording. Since you have high gain, recording in 8-bit is fine.

 

Stacking is a pita:

Surface mode in AS!2, and hope for ok image stabilization performance (I'm working on making this more robust, and I think I just found a bug that made it really tricky to get accurate surface alignment for deepsky imaging).

Typically a single alignment, and the other settings are often just lots of experimenting, just like processing the stack in Photoshop.

[Peter in Reno]

 

>Do you know approximately when this camera will be released?

I'm not sure really. I'll ask Sam.

 

Regarding the fixed pattern noise, what is the best way to quantify this? I mean, I can see one vertical line at high gain, but other than that, the picture looks quite clear. I can measure the standard devitiation of stacked bias frames (or maybe better: the stdev of many bias frames averaged, and subtract the read noise component). Assuming that fixed patter noise is what is what is left over in the images when there is no signal and the read noise has gone? But there must be good ways to look at columns and rows too??

I think the easiest way, just need to attach a stretched bias frame

224 bias.JPG

 

 

Hi Sam,

 

Can you post a sample image from another camera that has FPN to compare to ASI224? I do not know exactly what FPN should looks like.

 

Thanks,

Peter

[Vars191]

According to this web site, planned mass production of this new CMOS sensor is December 2015:

 

http://image-sensors...ity-sensor.html

 

That probably means cameras won't be ready till sometime in 2016?

 

Peter

 

That sucks, I found info on the sensor it is kick a*s   illumination of 0.005 lux or less

Edited by Vars191, 16 June 2015 - 09:34 AM.

[Vars191]

Hi Sam

Any idea of price? at this stage so I have a rough idea of how much I got to save to get one when it comes out can I pre order one now?

[ToxMan]

I'd like to see this on a SCT with Hyperstar.

[charotarguy]

The 16" Dobson was used, of course

 

>Can you also share the process of taking such images

Well, first I have to find the target, and typically I do this with Stellarium on my laptop followed by some starhopping (which is fun, but if you just stared on your laptop screen, it can be tricky). I use a 32mm 2" eye piece for this. M57 is easy to find of course, but some other deepsky targets can be much more tricky.

 

Then I switch the eye piece with the camera, move the focuser inwards - two turns or something like that - and I'm relatively close to focus (e.g., I typically then am able to see blobs of stars with 2 second exposures and high gain settings). I fine-tune the focus by switching to extremely short exposure times (typically I get 10fps) and paying close attention to one single star that is not over-exposed (so you can really see it 'shine' when it is perfectly in focus).

 

Then I put the camera settings correct (e.g. relatively high gain, but with this camera certainly not max. Just to have pretty much the best read noise performance), and start recording. Since you have high gain, recording in 8-bit is fine.

 

Stacking is a pita:

Surface mode in AS!2, and hope for ok image stabilization performance (I'm working on making this more robust, and I think I just found a bug that made it really tricky to get accurate surface alignment for deepsky imaging).

Typically a single alignment, and the other settings are often just lots of experimenting, just like processing the stack in Photoshop.

Thank you for the detailed information.