The Vela Supernova Remnant

[Itchy]

Originally Posted by PhotonCollector

Hi Itchy,

Param. Set 2 reduces the amount of in-camera sharpening and color enhancement.

Paul

Thanks Paul.

Covington's article talks about the parameter setting effecting the way the the EOS file Viewer and Photoshop convert the RAW frames. I wonder if RAW Linear Conversion with Imagesplus uses the parameter settings. I guess I should ask Mike Unsold!!

Cheers

[tornado33]

Good points.
If I open a raw image of a daytime scene in IRIS then convert it back to colour CFA, I get a colour image, but not a good looking one, as linera conversion with no automatic white balance isnt suited to daylight photography, leading to makers of digital cameras using their fancy in camera processing, to give us normal images.

I do wonder now, if we are better off WITH this processing, as we try to get astro images that as closely as possible mimic what the eye could see if it were many times more sensetive?

I have now found I get better eta carina images by opening the RAWs with the canon software (or in the case of my 300D with Photoshop CS), then stacking and processing them with either IRIS of PS Cs, as the nebula just seems to look better with less central burnout if done that way rather then letting IRIS do the RAW extraction via CFA, though for the really low surface brightness stuff Iris may be the go to try and lift the faint stuff fromthe sky background.
Scott

[Itchy]

Paul and Scott

I have just got word back from Mike Unsold (author of ImagePlus). He says (as I thought) that the digic camera settings have no effect on the image with Linear RAW conversion.

This is why I continually push my Linear barrow. If you want to convert your RAWs to achieve what the sensor recorded, you need to perform a linear conversion. If you convert to normal tiff (or Jpeg), the In-camera processing is going to do the stretching for you. The consequences are, as Paul has pointed out, the calibration will not be accurate. Parameter 2 may apply less processing, but it still mucks things up.

With ordinarly Linear conversion to tiff, the only processing that is applied is the Bayer colour interpolation and some white balance. This is OK, but there is still a better way.

What I now do exclusively is convert my RAW's to no white balance, pre-Bayer CFA (Colour Filter Array). Only after dark, flat and bias calibration are they converted to colour. This is the best way to get to the RAW data from the chip.

Cheers

[Itchy]

Originally Posted by PhotonCollector

That's right Scott,

Using Param Set 2, uses less in-camera colour manipulation and less sharpening of the image. But even with Param Set 2, the DGIC? processor in the camera still alters the original image data before you get to see it. It would be great is Canon had another Param Set which told the on-board image processor to just give us the image as is without any adjustment - then we'd be able to properly calibrate images and take astrometric measurements from them. Oh well - it's still a value for money astro-imager.

Paul

Sorry Paul, I just read this after I completed my post. This is exactly what Linear RAW conversion does. It bypasses the in camera processing (At least with ImagesPlus).

Cheers

[tornado33]

Hi
yes I believe IRIS does a similar thing, there is a Linear conversion box (ticked ) and the option to turn on or off colour balance. With Iris , nebulae seem to lack "punch" and outer detail, without burning in the brighter parts, but yes I do understand that Linear Conversion to pre bayer CFA monochrome images is the only way to accurately dark subtract astro images.What I wish was that I could restore the processed subtracted images to look exactly as they do if opened by the Canon software.

When processing a pic of Eta carina in Iris, the finished image is nice and free of dark current noise, but when I stretch the levels I seem to burn out the central parts while still not getting at all the outer faint detail. Images extracted from RAW with just the canon software show a lot more of this outer detail without burn in, I dont know why this is so. Perhaps Images Plus does a better job of the processing
Scott

[Itchy]

Hi Scott,

You should be able to achieve through processing what the digic software does to stretch an image. It is possible to bring out faint detail without saturating brighter areas. I usually achieve this using Photoshop Curves. The "punch" you are after should be achievable.

I'm not sure whether IP does a better job or not.

Cheers

[tornado33]

Thanks Tony
Im burning a CD of some of my Raw images as I type, along with some flats, an offset and darks, I will put it in the post tomorrow to you
Scott

[Itchy]

I look forward to getting them

Cheers

[h0ughy (David)]

SCott I now have IP2.75, its very complex but I think will give excellent results, will look into it sometime over the weekend

[tornado33]

Thanks Tony, the disc is in the mail.
Thanks Dave, yes will be good to see how IP goes
Scott

[PhotonCollector (Paul)]

Originally Posted by Itchy

Paul and Scott

I have just got word back from Mike Unsold (author of ImagePlus). He says (as I thought) that the digic camera settings have no effect on the image with Linear RAW conversion.

Hi Itchy,

I beg to differ, page 55 of my Canon Manual appears to say the opposite. It says "the image you capture can be processed automatically by the camera in accordance with the parameter settings.... Contrast, Sharpness, Saturation and Color Tone..."

Linear conversion has nothing to do with the camera parameter settings, it is two different subjects. The in-camera parameter settings are applied to RAW and JPEG images before they are stored on its memory card. This is easy to test as I did by doing some raw images with one Param Set turned up to Maximum and another Param Set to Minimum levels. You can see the diffence in the resulting RAW images when you load them (with Linear representation of the image). So I use Param Set 2 because it has all Param Settings set to zero.

Originally Posted by Itchy

This is why I continually push my Linear barrow. If you want to convert your RAWs to achieve what the sensor recorded, you need to perform a linear conversion. If you convert to normal tiff (or Jpeg), the In-camera processing is going to do the stretching for you. The consequences are, as Paul has pointed out, the calibration will not be accurate. Parameter 2 may apply less processing, but it still mucks things up.

With ordinarly Linear conversion to tiff, the only processing that is applied is the Bayer colour interpolation and some white balance. This is OK, but there is still a better way.

What I now do exclusively is convert my RAW's to no white balance, pre-Bayer CFA (Colour Filter Array). Only after dark, flat and bias calibration are they converted to colour. This is the best way to get to the RAW data from the chip.

Cheers

Linear conversion is what every astrophotographer uses ( I hope ) so that the intensity of light in the resulting image is directly proportional to the light that fell upon the sensor.

Where you say "If you convert to normal tiff (or Jpeg) the In-camera processing is going to do the stretching for you", again I do not agree with this. If you convert to Tiff/Jpeg then it is the image processing that does the stretching for you (not the camera).

In fact the camera will not stretch a histogram (even with JPEG images), instead it applies a Logarithmic stretch to the intensity of light that fell upon the sensor - this has the effect of "livening up" dull areas of a normal daylight image.

Anyhow, it's all fun and games eh?

Paul

[[1ponders] (Paul)]

Originally Posted by Itchy

What I now do exclusively is convert my RAW's to no white balance, pre-Bayer CFA (Colour Filter Array). Only after dark, flat and bias calibration are they converted to colour. This is the best way to get to the RAW data from the chip.

Cheers

Tony a quick question re your calibration process. Prior to doing your dark, flat and bias calibration are these files converted to nWB, pre-B CFA first or are do you just convert them to standard tiff prior to cal?

Cheers

[Itchy]

Originally Posted by [1ponders]

Tony a quick question re your calibration process. Prior to doing your dark, flat and bias calibration are these files converted to nWB, pre-B CFA first or are do you just convert them to standard tiff prior to cal?

Cheers

Hi Paul,

The files are converted to NWB pre baayer CFA first. After calibration, they are then "baayerised" to produce colour images.

Cheers

[Itchy]

Hi Paul,

It certainly is fun. I am also certainly up to being shown wrong on this one. Mike Unsold assured me that the parameter setting had no effect on Linear conversion of the RAWs, so I believed him. Your post has made me think again, so I decided to test it out. I took 10 x 10sec darks using Parameter 1 and 10 x 10sec using Parameter 2. Each set was taken alternatively to reduce the infulence of any temperature differences. I then converted each set using ImagesPlus with each of CWBlinear, CWBNonlinear and NWB CFA. The results are attached.

The in-camera parameter settings are applied to RAW and JPEG images before they are stored on its memory card.

This is the crucial thing that I was missing. Mike Unsold was also correct in that these setting had no effect on the RAW conversion. The effect was already in the RAW file.

Now looking at the results, there appears to be a difference between Parameter 1 and Parameter 2 regardless of the conversion type used. However, I applied a t test to the data and the differences bewteen the data can easliy be accounted for by random variation. From the stats you cannot guarantee that there is indeed a difference.

Having said that, I do conceed that there is a difference. If my assumption was correct, then the two sets would have been identical under linear conversion.

Interestingly, it appears that the noise is lower with Parameter 2 under linear conversion but higher with Parameter 2 under CFA.

As you said, fun and games.

[[1ponders] (Paul)]

Originally Posted by Itchy

Hi Paul,

The files are converted to NWB pre baayer CFA first. After calibration, they are then "baayerised" to produce colour images.

Cheers

Hi Tony

I just re-read my question and I don't think I worded it clearly enough 'cos now I'm not sure if your answer answers my question or not. Just bear with me.

Do you NWB and pre-B CFA your darks and flats first? This is what I was trying to ask.

[Itchy]

Originally Posted by [1ponders]

Hi Tony

I just re-read my question and I don't think I worded it clearly enough 'cos now I'm not sure if your answer answers my question or not. Just bear with me.

Do you NWB and pre-B CFA your darks and flats first? This is what I was trying to ask.

Hi Paul

Yes, Lights, darks, flats and bias are all converted using NWB CFA.