Monday, 5 October 2026

Exploring a ToupTek Astro AE676M monochrome multi-purpose astronomy camera. Part 1a: Fainter objects.

The reader is encouraged to re-read Part 1 before reading this Part 1a as some changes were made, to Part 1 post publication.

Further to the serious caviat about fixed noise in previews of captured images of the bright object M17 with 60s exposures; this test involved imaging a fainter object using 3 minute exposures.

One and a half hours of total exposure were captured by AstroDMx Capture of NGC7822, the central part of the Question mark nebula; half an hour's worth of 3 minute exposures through each of H-alpha, O3 and S2 Altair 2" 7nm narrowband filters, plus calibration frames using an Aurora electroluminescent flat panel for the flats.

The data were stacked and calibrated in PixInsight and further processed in PixInsight, GraXpert, SetiAstroSuitePro and GIMP3.

Click on an image to get a closer view

On the face of it, a reasonable image was constructed in the Hubble palette

NGC7822 Hubble palette


However as in Part 1, the imaging experience was not a pleasant one. The previews of the images were dominated by large amounts of fixed noise which completely obscured fainter parts of the nebulosity. The brighter parts of the nebulosity could be seen through the Ha filter but nothing of significance could be seen in the previews through the O3 and S2 filters, even though there was information present, as calibration and stacking later showed. This is unlike anything we have seen with other cameras, cooled or uncooled. To be clear, this fixed noise is not created by any preview transform, it is present in the raw FITS data.

To illustrate this point we used ASI DeepStack to stack quickly the Ha data and the S2 data with and without dark-frame correction. The resulting 16 bit stacked images were only stretched, 20% Bg 3 sigma in GraXpert.

Stack without dark-frame correction



Stack with dark-frame correction


Just dark-frame correction made a lot of difference removing most if not all of the fixed noise.

The problem is illustrated by the following two animations:

Animation blinking between an uncalibrated stack of the Ha data and the dark-frame corrected stack


The uncorrected stack shows some Ha nebulosity but some of it is obscred by the fixed noise.

Animation blinking between an uncalibrated stack of the S2 data and the dark-frame corrected stack


The uncorrected stack shows virtually no S2 nebulosity while some faint nebulosity is revealed in the corrected stack.

The argument here is not that it it is impossible to obtain images of faint nebulosity. The point being made is that the imaging experience is atypical and very poor with the live view showing little if anything in the fainter S2 and O3 data.

Further Tests in Solar imaging

The Sun in H-alpha light through a Coronado Solarmax II 60, BF 15 H-alpha scope. Captured by AstroDMx Capture with a Touptek Astro AE676C camera  in HDR 16 bit mode. The best 80% of 1000 frames stacked in Autostakkert!4, wavelet processed in waveSharp3 and further processed in PixInsight and GIMP3.


The Sun in white light through a William Optics Zenithstar SD Doublet 66mm APO refractor fitted with an ICE ND100000 glass solar filter. Captured by AstroDMx Capture with a Touptek Astro AE676C camera in 8 bit mode. The best 90% of 500 frame SER file were stacked in Autostakkert!4, wavelet processed in waveSharp3 and further processed in GIMP3.


There seem to be no problems with the camera on very bright objects, and in 8 bit or 16 bit HDR modes.

Thursday, 1 October 2026

Exploring a ToupTek Astro AE676M monochrome multi-purpose astronomy camera. Part 1.

The ToupTek Astro AE676M monochrome CMOS camera was attached to a Stella Mira 66mm ED APO  refractor which was fitted with a field flattener, an Altair V2 2" filter drawer, a camera rotator and a ZWO EAF focuser. A ToupTek GS-100, F=100mm, A=25mm quadruplet PAPO scope was mounted as a guide-scope with an Altair 462M2 monochrome CMOS camera as the guide camera as explained in the previous blog article . Autoguiding was done with PHD2 running on a Fedora mini computer via an INDI to control the Celesdtron AVX mount.

The equipment


The Altair 462M2 camera has an underpowered one stage TEK cooler with a detatchable fan unit that holds in place with strong magnets to enhance the weak cooling. The cooling is powered by the same USB cable that powers and streams data from the camera. Therefore we employed a 4 port SABRENT powered hub between the camera's USB-C to USB 3.0 cable and the USB 3 cable used to carry the signal indoors to the imaging computer. The cooling works but doesn't lower the temperature to those typically achieved by cooled astronomical cameras. Having said that, it did prove to be adequate for the job. It will clearly be better during the cold winter months. For solar imaging, the cooler helps prevent the camera from overheating.

The ToupTek Astro AE676M utilizes the Sony IMX676 sensor, which features a hardware architecture capable of a single-exposure dual-gain readout, often referred to in the industrial/surveillance sector as Clear HDR.

To understand how the HDR mode achieves a 16-bit output by combining HCG and LCG, it helps to break down the mechanics of the sensor's pixel design. 

Understanding LCG vs. HCG

In a typical CMOS sensor, every pixel acts as a tiny container storing light energy (photons) converted into electrical charges (electrons). The camera then converts these electrons into a digital number using an Analog-to-Digital Converter (ADC).

LCG (Low Conversion Gain): The sensor sets the pixel container to maximum capacity. This yields a high Full-Well Capacity (around 11,500 electrons for this sensor). It prevents bright targets like the core of a star, the moon, or the sun from instantly clipping into pure white. However, the trade-off is higher readout noise.

HCG (High Conversion Gain): The sensor electronically limits the container size but amplifies the signal intensely right at the pixel. This dramatically drops the Readout Noise (down to about 1.01 electrons), allowing you to see very faint signals buried in the dark background. The trade-off is that the pixel container overflows very quickly on bright details.

The HDR mode is intended for high-contrast targets: This mode  should be best for Solar, Lunar, and Planetary imaging, or highly uneven deep-sky objects like the Orion Nebula. In solar imaging, for example, it allows the capture of both the very bright solar surface and faint edge prominences simultaneously.

How HDR Mode Works (Single-Exposure Dual Gain)

The capture software must be set to capture 16 bit data for HDR mode to work properly.

Instead of forcing you to choose between the two, HDR Mode reads each pixel twice simultaneously from a single exposure using two independent readout paths on the sensor:

Simultaneous Capture: When the shutter closes, the camera reads the sensor through the HCG path (optimized for shadows) and the LCG path (optimized for highlights) at the exact same moment. Because this happens in a single exposure, it avoids the "ghosting" or motion blur artefacts caused by taking two separate exposures back-to-back.

On-Board Synthesis: The camera’s internal firmware compares the two readouts pixel by pixel:

• If a pixel is dark or faint, it extracts the data from the clean HCG readout.

• If a pixel is bright and saturating in HCG, it pulls the data from the high-capacity LCG readout.

16-bit Output Mapping: The native hardware ADC of the sensor is 12-bit. However, by stitching the shadow depth of HCG and the highlight headroom of LCG together, the total dynamic range spans past 12 bits. To accommodate this expanded range without truncating the data, the camera mathematically maps the blended information into a 16-bit container (storing values from 0 to 65,535). A number of camera manufacturers have this method available.

In this first part of exploring the Astro AE676M we only used the LCG and HCG modes and will reserve the HDR mode for future testing. The camera was fitted with a Baader UV/IR cut filter otherwise the image would be swamped by IR light to which the camera is very sensitive.

We used AstroDMx Capture to image the Aun in H-alpha light through a Coronado Solarmax II 60, BF 15 H-alpha scope mounted on a Skywatcher Solar Quest solar finding and tracking mount.

A 1000-frame SER file was captured in LCG mode in 8 bits. The best 80% of frames in the SER file were stacked in Autostakkert!4, wavelet processed in waveSharp3 and further processed in PixInsight and GIMP3.

Click on an image to get a closer view.

The Sun in H-alpha light


Even in 8 bits, the prominences could still be seen and captured.
Similarly we used AstroDMx Capture to capture a 1000-frame 8 bit SER file of the Moon through the Stella Mira 66mm ED APO  refractor fitted with a Pegasus L-UV/IR cut filter. The best 80% of frames in the SER file were stacked in Autostakkert!4, wavelet processed in waveSharp3 and further processed in GIMP3.

The 93% Moon



Deep Sky imaging in HCG mode

One and a half hours of total exposure were captured by AstroDMx Capture; half an hour's worth of 1 minute exposures through each of H-alpha, O3 and S2 Altair 2" 7nm narrowband filters, plus calibration frames using an Aurora electroluminescent flat panel for the flats.

Electroluminescent flat panel



The camera focused on a bright star using a Bahtinov mask



Screenshot of AstroDMx Capture capturing 60s exposures of M17 in H-alpha


The data were stacked in PixInsight and further processed in PixInsight, GraXpert, SetiAstroSuitePro and GIMP3.

M17 Hubble Palette



(H+S)OO L palette



Hubble - (H+S)OO L palette blend


HOS Canada, France, Hawaii Telescope palette



HSO palette



OHS palette


OSH palette



SOH palette


Pixelmath images were constructed using Siril

ForaaX palette



Gendler palette



Natural palette



In Part1 we have established that the Touptek Astro AE676M monochrome multi-purpose astronomy camera delivered good results using standard LCG and HCG modes. The results on a quite bright nebula, M17 with 60s exposures in HCG mode was good. Similarly in LCG mode on the Moon and on the Sun in H-alpha were good.

An adverse comment

Whilst the end results on the bright target M17 were quite good, the imaging experience itself was not good and was disappointing. The camera was set to the recommended values for gain and offset yet the preview images were full of huge amounts of fixed noise that didn't change as each new image came in. The noise was so intense that it obscured the fainter details that didn't become evident until after calibration and stacking. We did not use realtime dark-subtraction or live stacking. However, we have never seen this large amount of fixed noise from any camera, cooled or uncooled. This leads to the suspicion that there may be a fault on the camera that we have. We shall test on a fainter object, with longer exposures to determine what happens with respect to this fixed noise. In any event, we intend to contact Touptek to obtain their response. Touptek is arguably one of the very best astronomy camera manufacturers so this type of performance is atypical of their cameras in our experience.

We look forward to Part 2 in which we shall explore fainter nebulae with longer exposures in HGC mode and also HDR 16 bit  mode with solar and lunar imaging and also, when it becomes available, the Orion nebula.