Flea3 GigE focal length optimization question
Posted 08 June 2011 - 11:53 PM
Specifically, I have the Flea 3 with the Sony ICX 445 1 mb chip (pixel size of 3.45 um I think) in mono. I am running the PGR evaluaion trial - but of course had to buy the camera. I did not get the 618 only because I wanted to try something different with smaller pix and more resolution. I also just bought a manual Orion filter wheel and the Baader LRGB set of 1.25" filters, but have not tried anything but mono to date because of weather.
After a couple of weeks, I seem to have worked the kinks out of the video capture side (I think), am still learning Registax - but all of it is being made more difficult by a lack of simple knowledge with respect to what is the best way to fill this chip with my current scopes and equipment.
I have a C11 Edge HD that I typically image at f1.8 using Hyperstar and an M25C or Canon 5D, at f10 through the back and at f20 with a Baader VIP 2x Barlow. Initial attempts with this scope and the Flea3 seem very sensitive to seeing conditions.
I have therefore been doing most of my testing using my 5" ED 127 refractor (natural f7.5) and that same 2x Barlow (resultant f15 I assume). This seems fine, but the planet, even using ROI, is relatively small and pixelates quickly upon magnification after stacking.
Torsten has gotten FireCapture to work with the camera for me - which is great - but when I let the frame rate go as high as the ROI allows (I thought that was good!) - it can hit 70+. I suspect this high frame rate might be the cause of a post-processing ghosting effect (Saturn) in Registax.
To fill this chip up and slow down the FPS to, say, 30, I am thinking of trying EP projective as I have about every adapter imaginable at this point and figure I can set the distance to the chip on the Flea3 just about anywhere. I have a Baader 21 mm EP and a number of other Baader EP's that I would use.
Lots of info here from a newbie I know but I guess it boils down to a couple of questions:
- Is there a way to calculate how far I need to be from the chip using a 21 mm EP?
- What focal ratio would be optimal for each of my two scopes with this chip?
- Has anyone heard of high frame rates resulting in a ghost image/planet outline to the right of the object?
- How does one process 4 videos to LRGB in Registax? This is my first mono cam. Do you stack to 4 TIFF's and then stack the 4 TIFF's or do a PS4 post-process?
That's it for now. Really appreciate the help!
Posted 09 June 2011 - 12:07 AM
Posted 09 June 2011 - 01:19 AM
As for ghost images - you need to have some avis to check if they will be on everyone. There are problems with some cameras that constantly or randomly make such artifacts (not related to processing).
For RGB processing - I stack avis (usually castrator, autostakkert), and then process each in AstraImage (levels, deconvolution sharpening), and then I max RGB and XRGB (like IR channel for L) in this app (it asks for all the images and makes it).
Posted 09 June 2011 - 02:38 AM
pixel size* 5= FL
so 3.45 x 5= 17.25 rounded to a whole 17 !
As for the ghosting ???
More details re exp etc please.....
Posted 09 June 2011 - 07:25 AM
M = (s - Ef) / Ef
where s is the projection distance from the eyepiece to focal plane and Ef is the eyepiece focal length.
For your 21mm ep and a projection distance of 100mm you would have
M = (100-21)/21 = 3.76
So for your C11 that would give you approx a f/37 system. To get f/15 you would have to get the projection distance down to approx 50mm which is fairly difficult to do with a filterwheel in between. You can rearrange formula above to solve for distance:
s = Ef * (M+1)
for f/15 on a f/10 scope and 21mm Ep you get s = 21*(1.5+1) = 52.5mm
You would have to experiment at this point as the measurement of projection distance is subject to the uncertainty of where to measure from the Ep since it is a compound lens as well as determining where on the camera to measure to.
Posted 09 June 2011 - 01:27 PM
PiotrM, I have followed your posts for some time now as I have researched this end of the hobby - so thanks for replying. With respect to the appropriate f ratio to image using the ICX 445, below I have shown the f15 result without ROI (f7.5 refractor with 2x Barlow). Is that not too small a planet to work with vs. trying to fill the chip a little more? By the time I ROI into the planet to fill a frame, I am likely down into the 200x200 range tops. Also, thanks for the suggeston on LRGB processing. I assume I can accomplish the same things with PS4 (note I used levels and sharpen in the below pic with PS4).
CCS Hello, thanks and I will check out the link. Hopefully I can use it for my M25C as well as I have been trying to figure out how far from the Hyperstar optical surface I would need to a Pentax lens as well!
Karlo, exposures were set to 80% or so with gain and gamma down around 50-ish percent. The ghost only shows with a stretch and is most pronounced with ROI.
Desert Rat, I have been following your posts as well and appreciate the formulations! They raise another question for me: How far should the BARLOW optical surface be from the chip? My Baader VIP 2x allows me to break it down and go everywhere from putting the lens/nosepiece combo right into the filter wheel with no extension to adding probably 75 mm or so with the use of the full VIP spacers.
Bottom line with ghost problem: it is on every individual image takn as a TIF or any and all avi or ser video captures with a slight stretch of the image. Does this not RULE OUT this camera for me period and should I return the evaluation and get the 618?
Posted 09 June 2011 - 02:13 PM
Posted 09 June 2011 - 05:26 PM
Posted 09 June 2011 - 05:30 PM
Just check on at least few avis at max and almost max gain if the shadow is there.
Posted 09 June 2011 - 08:00 PM
Posted 09 June 2011 - 09:45 PM
How far should the BARLOW optical surface be from the chip?
For your camera I would start with the Baader barlow mounted directly to the filterwheel to keep the amplification down for the smaller pixels. Later you can extend it with T2 tubes when you need more amp for bright objects like Mars when seeing is better than average.
Posted 09 June 2011 - 11:47 PM
If you google a bit around you will see that you're not alone with this ghost artefact issue.
It seems to be a problem with all of these new ICX618-based cameras. The Flea3 (ICX618) seems
not having this (or only very faintly visible in some of the images shown here). The Basler Ace
has more problems with this. Here are some links related to this issue:
Basler Ace ghost problems
Basler Scout ghost images (german)
Allied Vision Guppy Pro ghost images
I would contact PGR support with this.
Posted 10 June 2011 - 02:09 AM
Mike what about ROI and binning , then throttling back the gain.
6.9 x 5= optimum FR of 34
But yeah I'd also be tempted to call PGR support
Posted 10 June 2011 - 02:38 AM
Karlo, thanks for the help. What is the 6.9x5 part though? Sorry for being a newb dumbas#.
ROI seems to increase the ghost issue. Binning, I have not tried. Isn't binning a decrease in resolution of the image (i.e. 2x2 is 1/2 of the resolution but still the entire 'size' of the frame?) I have not captured that way yet - but will try it. Would prefer to have a camera that does not have the issue so I am glad this was bought on a 'evaluation' with PGR. I have already logged a ticket with them on the topic - but they are uncharacteristically slow to respond (2 days old).
Now that I have this filter wheel and nice LRGB filters, even if I do switch, I plan on sticking with the mono experience. That's if they let me switch as the evalutation routine is very specific about what you can and cannot complain about.
I am going to remove the nosepiece of my 2x barlow (the part that contains the optical surface) and hook it straight to the filter wheel and try binning this weekend - hopefully from a dark site.
I spent most of the night tonight trying to figure out how to process LRGB. I have gotten pretty good with PS4 and my larger chips on DSO's, but this is a totally new animal and I can't really figure out how to do it with Registax 6 - which is finicky. Any tips or links to a step-by-step would be great!
Posted 10 June 2011 - 03:01 AM
Posted 10 June 2011 - 05:12 AM
3.45 pixels great for Luna/solar but a tad tincy for planetary IMO, but if you bin 2x2 then yes you're increasing the pixel x4 and losing some res, but that's okay for planetary, then using the formula I mentioned earlier from another thread, the optimum FR for the 2x2 pixels would be 34. You could then throttle back the gain quite a bit and see if any improvements occur- just an idea !
Posted 10 June 2011 - 08:06 AM
Keeping fingers crossed.
Posted 16 June 2011 - 02:46 PM
I have tested the new FW version with the ICX 618 today using a Pentax lens with both FlyCapture and FireCapture. I note that in flycap 1 or 2, the requested FPS and processed FPS reads at 120 with no binning – but the DISPLAYED FPS reads at 60 mhz (the max speed of my laptop monitor). Firecapture only shows me 60 mhz. PGR told me that the camera, even though it is displaying only 60 Hz will be capturing in 120. Is that the case with Firecapture and if not, are there any setting changes I need to effect to get it to 120 fps?
Lastly, I read somewhere that people are using this camera (ICX 618 based) and FireCapture to do longer exposures (26 sec)? Is that the case, if so, how do I go about doing that! I have a strong interest in DSO work with this camera as well.
Torsten, thanks again for the wonderful, custom support!
Posted 16 June 2011 - 08:01 PM
Mr. Willems, the Flea3 GigE is intended to come out in a model that uses the ICX618, but it is not out yet. Word I get from PGR is Q4. The model I tried was the first GigE and had the ICX 445 in it. It has admitted ghosting/bleeding issues with even subtle stretching. I could not run any significant wavelet or sharpen filter on the above photo without what looked like either a ghost image to the right or almost like water stains / bleeding in the object coming through. Then again, maybe I don't know what I am doing but engineering guys at PGR in Germany picked up one of my messages one day, tried it themselves nd confirmed the issue - which is now purported to be with engineering.
Posted 17 June 2011 - 03:06 AM
Posted 17 June 2011 - 03:21 AM
Posted 08 May 2013 - 12:02 AM