94 replies to this topic

[cuivienor]

This is for information: I've bought an ASI294MC Pro back in August last year, and have had decent success with it from darker skies (Bortle 5 or better) - OSC from Tokyo however was really not giving me anything.

 

So after seeing tons of reports about how good the OPT Triad Ultra filter is (and how others use it without issue with the ASI294MC Pro), I bought it to use from Tokyo with my ASI294MC Pro.

 

But unfortunately, I found that all my images with the filter came out with a very pronounced X shape pattern across the whole sensor. It is from the sensor. And in multiple tests to identify the origin, I found the same issue with Astrodon HA and SII filters (but not OIII). So the sensor exhibits a X shape sensitivity unevenness in deep red wavelengths.

 

Unfortunately, while flats happily show the patter as well, it doesn't calibrate out properly. It calibrates out the light intensity, but the red cast is still there, even in calibrated frames (some examples here: https://drive.google..._4mqxSnaLkmq2PN). This means I have to use cropping and aggressive DBE to get anything out of the frame, which is extremely frustrating.

 

Upon discussing that with ZWO support, they told me the manufacturer of the sensor said that the issue is within tolerance, so that they will not repair or replace the camera under warranty.

 

Obviously this is very disappointing. Especially since others with the same camera and filter do not experience the issue.

 

So word of advice: if you have the ASI294MC Pro and want to use it with OPT Triad filter, first borrow an HA filter from someone and get some flats, see if you get a terrible pattern. If you do, well, don't buy the OPT Triad filter just yet...

 

[ngc7319_20]

Are you taking separate flats in each color filter?

[cuivienor]

It's an OSC camera, with the OPT Triad being a multi-band narrowband filter for OSC. So there is a single filter. Flats and lights with the same filter.

I tried with my Astrodons just to be absolutely certain that cross shaped thing across the sensor was not the OPT Triad filter's fault...

[jdupton]

Yannick,

 

   Pardon my laziness for not searching out your other thread. I seem to recall that you did your testing with DSLR lenses. Is that correct? I don't recall you trying the Triad or Astrodon NB filters on anything slower than F/4 although I may have missed or forgotten it.

 

   What is the bandpass verson of the Astrodon filters you tried that failed (Ha and Sii)? I know the Triad has pretty narrow bandpass windows. If your Astrodon filters are of the 3nm series, it makes me wonder if the sensor cover glass on the 294 chip may be thicker than other cameras or multicoated. I know some very narrow bandwidth filters can cause problems at low focal ratios. The edges of the chip experience passband shifts and extinctions such that the pattern may resemble a cross. I wonder if perhaps the cover glass over the sensor were thicker, whether it would worsen that issue.

 

   If you do have the very narrow bandpass Astrodon filters, do you have other wider filters or someone near you that could loan you a 5nm or wider variety to try. (Your Oiii filter working is what triggered this thought for me.) If the issue you are encountering is a bandpass issue being made worse by the camera, then others could factor that into their decision making process.

 

 

John

[cuivienor]

Thanks John!

 

Yes you are right, the F ratio being used is quite important!

 

This is something I considered - so I tested with three different lenses:

- EdgeHD 800 with focal reducer, F7

- Kowa Prominar 350mm F4 and F4 and various slower speeds (using the diaphragm)

- Samyang 135mm f2, at various speeds, including F11 (using the diaphragm)

Still, I have also seen people using the OPT Triad Ultra with the RASA8 at F2 and ASI294MC Pro, again without issues.

 

My Astrodons are 3nm bandwidth and unfortunately, I don't have less narrow than that.

 

However I do love your theory - if it's the glass in front of the sensor... I guess it could be that my sensor has a thicker glass than others? In which case it could be affected even at slow focal ratios?

 

Can you think of any further tests I could try doing? 

 

Thanks again for the feedback! The last thing I want is to mislead people...

Edited by cuivienor, 18 May 2019 - 12:27 AM.

[ChrisWhite]

Yannick, I have seen your work and know you are a capable processor, so my instinct is to rule out pre-processing, however I looked at your files and I think that it's worth visiting flats again. 

 

How are you taking your flats?  Seeing that the pattern exists in the flat frames gives hope that it can be calibrated out.  It looks like there is a mis-match though with the flat.  The upper left is over-correcting fro example.  I'm not a user of the 294, but I know that with my CMOS experience (1600, 183) that flats are tricky and if they are not executed just perfectly they don't always work.

 

Another question, how are you calibrating?  Manually or with BPP?  if manually, what are your settings?  Share screenshots of your calibration window if you dont mind. 

 

Lastly, I remember a thread from a while back from Andy where there were strange gradients due to uneven cooling.  If you really push the data when you are not using an NB or Triad filter, are you able to see the gradient?   Wondering if it is there in a standard OSC shot, but that there is so much signal it buries the gradient. 

[cuivienor]

Thanks Chris!

 

I would indeed love it if I could find a way to calibrate this out (while still being annoyed that my sample of the camera has the issue, while other users of the same filter don't ). Preprocessing is one of the areas where I take a more relaxed approach.

 

So on the flat, I tried multiple methods, but unfortunately didn't keep everything (I threw away some I had taken with great care after the imaging session, because seeing the end result, I thought they were the cause of the X...). The one I shared was indeed not taken in ideal conditions, as it was white t-shirt + white wall, but in a fairly bright room, so light leakage was unavoidable. The one thing that makes me think the calibration is not that bad is that when taking the luminance of the calibrated frame, the X doesn't appear - so in terms of light intensity, calibration seems to be doing its job (except for that top left -> bottom right gradient in this example of the flats, which is still visible in the luminance).

 

At any rate, I tried with BPP, manually, and with DSS (with multiple settings). When doing it manually, I typically follow the steps of the light vortex astro pre-processing guide (but no bias, no cosmetic correction, without dark frame optimization, and plus debayering after calibration). I'm attaching my calibration screen and my flat stacking screen. Also tried various other options (median vs average, etc.).

 

As for the cooling - I had the same thought (and so did ZWO Support), so I tried without any cooling. Cooler off, small deltas, larger deltas. Short exposures, long exposures for the flats. Even very short exposures (a few milliseconds with flat, Sharpcap live view). The X is there, unchanging, extremely stubborn.

 

In a standard OSC shot, or even with a red filter on (yes, I tried with a broadband Optolong red filter), no matter how I stretch the data, I cannot see the X, even if I squint very hard... I can't really see gradients either.

 

If you have suggestions for better calibration (I have taken and tried better flats than the ones I shared, but I think I threw them away in frustration each time), I will gladly try!

 

Edit: as soon as I have some time, I will try to take better flats, will get back once I do!

Attached Thumbnails

• 
• 

Edited by cuivienor, 18 May 2019 - 08:39 AM.

[ChrisWhite]

If you dont mind, please share a dozen of each frame with me.  Lights, Flats, Darks, Bias.  I'd like everything .fit from the camera, without any processing.  If you do I am happy to try with my workflow and settings to see if I replicate the issue.

 

What is your light source fort the t-shirt flats? 

 

I struggled with flats for a long time with the ASI 1600.  The wall flats method you are using seemed to do the best for me, but eventually I went to a Spike-a-flat panel and that gives me perfect flats everytime.  That said, there may be hope that flats can remove it if you can get a perfectly matched flat, however I am inclined to doubt that will be your solution.

 

I'd like to look at the data a little more closely, but if you do decide that it is the camera I would push back at ZWO that this is not normal behavior.  Sure the sensor might be within normal manufacturing specs for terrestrial photography, but that is not the application you are using it in and not the application that ZWO has marketed it in.  While it is not super common to use narrowband with OSC, there should be nothing like this preventing you from doing so.  Very interesting to me that it is only in Ha and Sii and that boradband and Oiii do not exhibit the issue. 

[CharlesW]

Since you can’t apparently use any filter with that particular camera, wouldn’t it be more accurate to say, if you live in one of the most light polluted cities on the planet, don’t get the ASI294MC Pro?

[jgraham]

It is always good to be cautious. It would be interesting to hear if anyone else has seen something like this. I ended up with a pair of ASI294MC Pro's and I have not had any problems with the Triad on my Comet Catcher (5.5" f/3.5 Schmidt Newtonian). 

 

 

Alnitak Nebula Complex in Orion – Triad Filter First-Light
Telescope: Celestron Comet Catcher 5.5” f/3.6 Schmidt Newtonian, Orion Atlas EQ-G
Camera: ZWO ASI294MC Pro, 0C; Gain: 200
Filter: 1.25” OPT Triad Hb, OIII, Ha filter
Guide scope: Williams Optics 50mm, Meade DSI Pro II, PHD
Exposure: 20x240sec saved as FITS
Darks: 32x240sec saved as FITS
Flats: 32x0.25sec Sky flats taken at dusk
Average Light Pollution: Red zone, poor transparency, high humidity
Lensed Sky Quality Meter: 18.0-18.3 mag/arc-sec^2
Stacking: Mean with a 2-sigma clip.
White Balance: Nebulosity Automatic
Software: Nebulosity, Deep Sky Stacker, Photoshop

 

My sky conditions are certainly not as bad a Tokyo, but they t'ain't great either (Bortle 8). I often have to deal with oddball gradients due to my sky conditions which can be quite variable during an imaging run. In that case I use a synthetic flat, which does a great job reducing color gradients. In the example shown above I just used conventional flats.

 

It's always something...

 

 

[cuivienor]

@ChrisWhite thanks you for the offer! I will do additional, more careful flats with wall, wall+t-shirt, and my flat field box. I will send that across to you, along with the lights I took of two targets (Eagle and Omega Nebulae). Give me a couple of days to do that! As for ZWO answer, yeah to be honest I was very surprised - they were very apologetic and suggested some ideas with how to apply the flats, but wouldn't agree that having this huge X in the first place isn't normal. To be fair, I've been contacted separately by a ZWO representative after this post, so maybe with their help I can get to the bottom of this!

@CharlesW: that could be one way to put it! But then I think it's clear that my camera sample is different than others who have done narrowband with it (seen several here and on the ZWO FB Group)... I'm almost tempted to buy a second sample of the camera, see if I get the issue, and if not just sell the first one, being very clear about its limitation.

@John: thank you for your help separately! Yes your images with the filter and camera along with others using the same filter and camera (some using RASA8) convinced me to get it! Now if only that X wasn't there... It's always something indeed...

Edited by cuivienor, 18 May 2019 - 05:20 PM.

[cuivienor]

Well I was all set to take flats, but had another idea first. I fitted the filter on my Panasonic Lumix LX100m2 compact camera and took some flats, just to double check - No pattern at all, everything fine.

 

Then I took pictures of the sensor itself, through the filter. Here is one.

 

 

Besides the funky colors, looks good right?

 

But look closer. Take the red channel. Stretch it. Notice anything weird?

 

 

How about on this picture? (actually a stack to increase SNR - Stacks are not just for astro)

 

 

Oh and as a reminder here is what a debayered and stretched flat looks like (for demo purposes):

 

 

Does it look familiar?

 

Seriously this is a defective sensor. If it isn't then I don't know what is.

@ChrisWhite: Sorry this interrupted me, I'll take the flats later. But really all I want now is for ZWO to do something about this camera.

[ChrisWhite]

Um, I agree. I'm not sure what to make of it, but there does seem to be something wrong with that sensor.

It's clearly not the filter. If it was the entire image would still have a large "x". This is clearly isolated on the sensor itself.

Don't bother with flats at this point. I'd show this to ZWO and hopefully they will respond with a solution.

I've never seen anything like that before. So strange that it only impacts red....

[ChrisWhite]

Additionally, I bet if you take an broadband image without any filters and separate the channels you might see that patter buried in red.

[cuivienor]

Quick update: after talking more ZWO, they agreed I can send back the camera to them for inspection. I will do so, going through my local Japanese reseller, as per due process.

[ChrisWhite]

Quick update: after talking more ZWO, they agreed I can send back the camera to them for inspection. I will do so, going through my local Japanese reseller, as per due process.

Good to hear, and good luck.

[JP50515]

Quick update: after talking more ZWO, they agreed I can send back the camera to them for inspection. I will do so, going through my local Japanese reseller, as per due process.

I've been lurking your posts for a few weeks now. Good on you for digging into this and coming up with some hard evidence of a defect. I hope ZWO stands by their product here and replaces the unit, I know it's been driving you up the wall for some time now 

[cuivienor]

Well, there's bad news. Sam from ZWO examined the camera, and with my instructions (take flats through an HA or SII filter) managed to reproduce the issue.

 

But they tested other cameras and:

- All the ASI294MC Pro they tested had a pattern of some sort, of a different shape than mine. I asked them to send me some sample flats from other 294MC Pro, because I find that hard to believe knowing others have had success with the 294 and narrowband
- All the ASI1600MC they tested had the pattern, but worse

- ASI183MC Pro and ASI071MC Pro do not exhibit the issue.

 

So it seems to be a sensor "issue", maybe as written previously in this thread the Bayer matrix reacting weirdly to certain deep red frequencies.

 

They will be looking for good ways to calibrate that out.

Edited by cuivienor, 13 June 2019 - 05:32 PM.

[ChrisWhite]

Interesting report.  Please report back if they figure something out.  I don't have or intend to get any of these cameras, but interested in the resolution. 

[bobzeq25]

Well, there's bad news. Sam from ZWO examined the camera, and with my instructions (take flats through an HA or SII filter) managed to reproduce the issue.

 

But they tested other cameras and:

- All the ASI294MC Pro they tested had a pattern of some sort, of a different shape than mine. I asked them to send me some sample flats from other 294MC Pro, because I find that hard to believe knowing others have had success with the 294 and narrowband
- All the ASI1600MC they tested had the pattern, but worse

- ASI183MC Pro and ASI071MC Pro do not exhibit the issue.

 

So it seems to be a sensor "issue", maybe as written previously in this thread the Bayer matrix reacting weirdly to certain deep red frequencies.

 

They will be looking for good ways to calibrate that out.

Might they let you give them $, and trade it for an 071?

[cuivienor]

@ChrisWhite: yes I will update. It's just something to be aware of apparently: some Bayer matrices may be prone to exhibit a pattern in red narrowband.

@bobzeq25: actually in the meantime, I found a like new 071MC Pro for less than 1000 dollars, so jumped on it. It's happily gathering data with the Triad Ultra on the Crescent Nebula :-) But it's a good idea - since it's the same price, I'm asking whether I could exchange for a 183MM Pro. I'd then just buy narrowband filters from ZWO so I have two pure NB setups at home, and use OSC on the go. A bit of a stretch, but it doesn't hurt to try.

[Ballyhoo]

This is for information: I've bought an ASI294MC Pro back in August last year, and have had decent success with it from darker skies (Bortle 5 or better) - OSC from Tokyo however was really not giving me anything.

 

So after seeing tons of reports about how good the OPT Triad Ultra filter is (and how others use it without issue with the ASI294MC Pro), I bought it to use from Tokyo with my ASI294MC Pro.

 

But unfortunately, I found that all my images with the filter came out with a very pronounced X shape pattern across the whole sensor. It is from the sensor. And in multiple tests to identify the origin, I found the same issue with Astrodon HA and SII filters (but not OIII). So the sensor exhibits a X shape sensitivity unevenness in deep red wavelengths.

 

Unfortunately, while flats happily show the patter as well, it doesn't calibrate out properly. It calibrates out the light intensity, but the red cast is still there, even in calibrated frames (some examples here: https://drive.google..._4mqxSnaLkmq2PN). This means I have to use cropping and aggressive DBE to get anything out of the frame, which is extremely frustrating.

 

Upon discussing that with ZWO support, they told me the manufacturer of the sensor said that the issue is within tolerance, so that they will not repair or replace the camera under warranty.

 

Obviously this is very disappointing. Especially since others with the same camera and filter do not experience the issue.

 

So word of advice: if you have the ASI294MC Pro and want to use it with OPT Triad filter, first borrow an HA filter from someone and get some flats, see if you get a terrible pattern. If you do, well, don't buy the OPT Triad filter just yet...

I would not think to image with the 294 from the most densely populated city in the world. But wow you did. 

[17.5Dob]

Might they let you give them $, and trade it for an 071?

I was already  to bite on the cheap 294, several months ago, but the more I read, the worse it sounded. I guess the adage is still true.."You get what you pay for in AP "

I'll just keep muddling around with my uncooled, 247C aka a D5300

Edited by 17.5Dob, 13 June 2019 - 11:15 PM.

[cuivienor]

@Ballyhoo: to be fair, I'm a bit crazy I'm now using the 071MC Pro instead, and it's getting good results on the Crescent Nebula

 

@17.5Dob: yeah, the 294 is actually quite good for its intended purpose: OSC in low light pollution (red zone or better - yeah, red zone is low LP for me). Very fun to watch the subs come in so quickly. And very long stacking times, you'll want to invest in a multi-core CPU and install Linux + PI Linux for best performance But it has to be used at high gains (including for flats), and it can feel like you're tinkering with it from time to time. In that respect, the 071MC Pro is very stable, if less sensitive.

[andysea]

This confirms my observations on the qhy294.

I was getting a weird red pattern and when I looked at the sensor under diffused LED lighting in my office I could see a mottling that matched the artifact that I was getting in my images. It also confirms my hypothesis that there is a variation from sensor to sensor and the bad ones are non calibrateable.

I took my qhy 294 back to the store so have qhy look at it. If I get it back I will definitely run your test. I don’t have an OSC narrowband filter butt I will try with an astrodon 3nm or 5nm filter and monochrome camera.