Use AI to help solve general imaging problems

[oopfan]

Conventional search engine requests return a ranked list of relevant results, and depending on the query you can find your answer, however, it takes effort. It behooves you to visit the top 5 results and judge for yourself. Many times you can find what you need but like I said it takes effort. Now imagine a technology that automatically looks at the top results and then summarizes what it found in a human-readable form.

My favorite AI search tool is https://www.phind.com which advertises itself as "Your personal search-enabled coding agent." I am a software developer and Phind has saved me countless hours of research the conventional way. So I was curious, can it help with general astro imaging problems?

A forum member recently asked a question about his noisy stacked image. Robin, who is on vacation, was kind enough to reply. So, I plugged the OP's question into Phind. I was quite amazed but certainly not surprised by the quality answer:

To identify the cause of the excessive noise in your captured image and determine if the issue lies with SharpCap or StarTools, we need to investigate the possible factors that could contribute to the noise. Here are the steps you can follow:

1. Check Image Stacking: Noise can often be reduced by stacking multiple images. Make sure you have enough light frames, preferably with longer exposure times, to improve the signal-to-noise ratio. Consider capturing more light frames to increase the overall exposure time.
2. Evaluate Calibration Frames: Calibration frames such as darks, flats, and bias frames are essential for noise reduction and calibration in astrophotography. For dark frames, ensure that they were captured at the same temperature and duration as your light frames. Variations in temperature or exposure time can lead to noise issues. Additionally, check if you correctly applied the master dark frame during image processing.
3. Assess Flat Frames: Flat frames are used to correct for uneven illumination and dust on your camera sensor. Make sure your flat frames are properly captured and applied during processing. If you did not include flats in your initial processing, try adding them to see if it improves the noise issue.
4. Examine Camera Settings: Check your camera settings to ensure they are optimal for capturing images. These include gain, offset, and temperature settings. Higher gain values can increase noise, so try reducing the gain if it's set too high. Similarly, make sure the offset value is at an appropriate level.
5. Evaluate Software Settings: Review the settings within SharpCap and StarTools to ensure they are properly configured for image processing. While both software are widely used and known to be reliable, it's worth examining their settings to rule out any potential causes.
6. Experiment with Different Software: To determine if the issue is related to SharpCap or StarTools, try processing your images with alternative software such as PixInsight or DeepSkyStacker. This will help narrow down the potential sources of the noise.

Give it a try!

Brian

[admin]

Hi Brian,

well, that is interesting... The cynic in me is however wondering if somewhere here on the forums (or on Cloudy Nights, or Stargazers Lounge or somewhere) there is an answer to a very similar question that is basically a paraphrase of the above. That will be in the training data of the LLM used for the AI search and therefore the above answer comes out.

Can you ask it to give references that support the answer it gives - that would be useful!

cheers,

Robin

[oopfan]

Hi Robin,

The model that Phind uses is updated frequently, that's why it is preferred over ChatGPT, the free-to-use version. Having said that this morning, after reading your reply, I posed the same question to Phind as the day before. Its reply was different but still useful although not as good as the day before. However, I did take note of the footnotes. There were several links to SharpCap forum posts, one to StackOverflow, and some others that I did not recognize. Generally, there are 8 or more citations in every Phind result that I've made. I'll try again tomorrow and the day after.

Brian

[oopfan]

Robin,

I ran the same query again this morning. Like yesterday's results it was "ok" but not as good as the day before. I know one thing, Phind is undergoing rapid development. The UI changed significantly the last couple days, and who knows what else.

This morning I noticed that two checkboxes were unchecked: "Use Best Model (30 left)" and "Legacy Search". I tried checking "Use Best Model" but it wanted me to log in. I don't have an account so I aborted that, and then tried "Legacy Search". That worked. Interestingly, it gave me a result very similar to the other day. It cited sources but nothing eye-opening. Mostly Wikipedia links to the Fireworks Galaxy. Then I copied portions of the results and input it into a regular Google search. It found nothing that makes me think that Phind extracted it from one particular source. It really does appear as if Phind is ingesting the Web and synthesizing an answer based on the highest ranking results.

Here is what it gave me today:

The query I gave it:

I am having some real difficulty imaging the fireworks galaxy (NGC6946) and was wondering if you could assist. I took about 4 hrs of 2-min subs last night under what appeared to be ideal conditions, but there was excessive noise in the image which I couldn't remove when processing it (using startools). I am not sure if the noise issues were with startools or on the capture (sharpcap) side. When the raw stack (using dss) included lights, darks, and flats, stars were present but the image was not visible at all when I went to process it. When I removed the darks from the raw stack the image appeared when processing, which tells me there was something wrong with my darks, so I removed them from the raw stack. I then tried to process it using only lights and flats, and the image appeared, but there was excessive noise. The noise was so severe I couldn't process it any further. Can you assist in identifying the cause of the noise? Is it on the sharpcap or startools side?

Its reply:

ANSWER | GPT-3.5 MODEL

To assist with identifying the cause of the noise in your image of the Fireworks Galaxy (NGC 6946), we need to analyze the steps you took during capture and processing. Based on the information you provided, it seems that the noise issue may be related to the dark frames used in your stack. Let's break down the steps and address each one:

1. Capture: You mentioned using SharpCap for capture. SharpCap is a popular software for astrophotography, and it offers various settings for image capture. It's essential to ensure that you have set the correct exposure time, gain, and temperature (if applicable) for your camera. These parameters can affect the amount of noise in your images. Additionally, make sure your camera sensor is clean and free from any dust particles that could contribute to noise.
2. Calibration frames: You mentioned using lights, darks, and flats for calibration. Calibration frames, such as darks and flats, are crucial for reducing noise and correcting for imperfections in your images. Dark frames are used to subtract the camera's noise pattern from the lights, while flat frames correct for uneven illumination. However, if the dark frames are not properly matched to the lights in terms of exposure time and temperature, they can introduce additional noise into the image. It's important to capture dark frames with the same exposure time and temperature as your lights. You may need to retake your dark frames to ensure they are properly matched.
3. Stacking: You mentioned using DeepSkyStacker (DSS) for stacking your images. DSS is a popular software for aligning and combining multiple images into a single stacked image. When stacking, DSS uses the calibration frames (darks, flats, and bias frames) to improve the overall quality of the final image. However, if the dark frames are not properly matched to the lights, or if there are issues with the calibration frames themselves, it can lead to noise in the stacked image. Make sure to double-check the settings in DSS and ensure that the calibration frames are being applied correctly.
4. Processing: You mentioned using StarTools for processing the stacked image. StarTools is a powerful software for astrophotography image processing. It offers various tools and algorithms to enhance and optimize your images. If you're experiencing excessive noise during processing, it could be due to aggressive noise reduction settings or improper use of the software. Experiment with different noise reduction settings in StarTools to find the right balance between noise reduction and preserving details in your image. It's also worth considering using other image processing software, such as PixInsight or Adobe Photoshop, to see if they yield better results.

To summarize, here are some possible causes of the noise in your image and steps you can take to address them:

• Ensure that your camera settings (exposure time, gain, temperature) are appropriate for capturing the Fireworks Galaxy.
• Check your camera sensor for dust particles and clean it if necessary.
• Capture dark frames with the same exposure time and temperature as your lights to ensure proper matching during calibration.
• Double-check the settings in DSS to ensure that the calibration frames are being applied correctly during stacking.
• Experiment with different noise reduction settings in StarTools or try using other image processing software to see if they yield better results.

Remember that astrophotography is a complex and iterative process, and it often requires experimentation and fine-tuning to achieve the desired results. Don't hesitate to try different approaches and seek guidance from the astrophotography community or online forums for further assistance.

Brian

[admin]

Hi Brian,

I had a play on that web site later on after posting, and also noticed that it does tend to quote references for where the information comes from, which is useful. Also interesting that while it is really claiming to help with programming questions, it knows all about *using* various bits of software too.

A useful resource - thanks for pointing it out as it may short cut the sort of 'digging round the web collecting info on a problem' that often takes time.

cheers,

Robin