Mindat Logo
bannerbannerbannerbanner
Welcome!

Amscope MU1000 camera - initial impressions

Posted by Henry Barwood  
Amscope MU1000 camera - initial impressions
November 02, 2011 06:54PM
us    
I received my Amscope MU1000 camera today and thought i would post some initial impressions of the camera.

Positive:

The software istalled easily on my W7 platform. When i hooked up the USB camera, the drivers installed properly and functioned as they were supposed to. This is not always the case with Chinese software.

The program ToupView is surprisingly easy to use and (once you locate them) has adjustments for White Balance and exposure settings. The WB is a bit odd as it is a sliding scale that adjusts the color temperature. This does allow fine tuning of the color of your specimen.

Live view is good, but occasionally flares out as brightness changes.

Shutterless, so very minimal vibration effects.

Negative:

While advertised as a 6.1 megapixel camera, it is not. The basic chip is 800 X 600 pixels. The larger value is derived from software interpolation and actually gives lower resolution than a typical HD webcam!

The negative lens that projects the image to the chip suffers from severe radial astigmatism. This is very bothersome with higher magnification images.

There is no way to adjust the camera for differing magnifications. What you have on your scope objective is what you get. In my case 0.66 to 4 X.

Overall:

Not bad, but not good. You can get a lot more resolution for the same money with other cameras. It is quite easy to use and lacks shutter vibration problems, but with the minimal resolution this is not a significant factor.

Henry Barwood
Troy University
Troy, Alabama USA
avatar Re: Amscope MU1000 camera - initial impressions
November 06, 2011 12:34AM
us    
Hi Henry,

I’m confused and not sure what to think about your comments regarding the AmScope camera. I checked the specifications on the AmScope website and found the following.

Sensor Size 6.119mm (H) x 4.589mm (V) 7.649mm (diagonal)
Optical Format 1/2.3-inch (4:3)
Pixel Size 1.67um x 1.67um

Considering the horizontal width of the sensor, 6.119mm, and dividing by the pixel size, 1.67um, I get 6119/1.67 = 3664 horizontal resolution. For the vertical it is 4589/1.67 = 2748. The resolution is 3664X2748 = 10,068,670 pixels, so 10MP sounds about right. This is also verified by the sensor manufacturer’s specifications as well. [www.aptina.com] Therefore, I am curious as to how you determined that the camera uses interpolation to achieve higher resolution? I have used the several of the cameras in this series, under a different brand name, and have not found the specifications to be incorrect.

What I have found to be detrimental to the performance of this camera, for my purposes, is the small pixel size. 1.67um/pixel is very small and from the standpoint of well depth, and considering dynamic range and noise, larger pixels would be preferred. On my microscope, a Nikon SMZ800, this camera is not a good match for the resolution of the scope and probably isn’t for yours as well. My objective is a Nikon ED Plan 1X, which has an NA of 0.09 or a calculated resolution of about 3.7um on the subject. Actually, I have measured it to be a bit better. Since my field width at 63X, is ~ 2mm wide, the resolution on the sensor ~2000um/3664pixels ~ 0.54 um/pixel. Usually a sample of 2 pixels is required to satisfy the Nyquist sampling theorem, but because of the Bayer pattern, 2.5 is considered to be more accurate. For idea sampling then, 3.7um/2.5 pixels = 1.48um/pixel. This is almost three times larger than the resolution at the sensor of ~0.54um/pixel. The image would be oversampled by a factor of ~3. Unfortunately, the binning to 1832X1374 is only 2.5MP, so that is under sampled and doesn’t match well either. So, I believe that a 5MP camera would provide about the same image resolution, on my scope, as a 10MP camera. The 5MP camera’s larger pixels would also provide greater dynamic range and lower noise. The 10MP camera would be of more benefit, and pixels would not be wasted, on a compound microscope with higher NA objectives.

So, more megapixels aren’t always a good thing, leading me to wonder about today’s sensors that now exceed 120MP! [www.dpreview.com]

Gene
avatar Re: Amscope MU1000 camera - initial impressions
November 06, 2011 07:50PM
us    
Gene,

In your calculation of pixel density, did you factor in the three colors, RGB, and their area fill efficeincy? This could reduce the effective pixel density by a factor of as much as 10.

Ron
Re: Amscope MU1000 camera - initial impressions
November 06, 2011 10:14PM
us    
Gene,

I take a look at a sharp edge (crystal or grain in a thin section) from low to high resolution. If all I get is increased noise, the the sensor is operating at the lowest resolution.

Ron,

I was going to suggest that, but you beat me to it.

Henry Barwood
Troy University
Troy, Alabama USA
avatar Re: Amscope MU1000 camera - initial impressions
November 07, 2011 02:26AM
us    
Henry and Ron,

As I mentioned, if the image is oversampled, then you are not using the full resolution of the sensor. Another way of looking at the problem is to assume for a moment that the resolution of the scope is 5um. If the resolution on the sensor is 0.5um then you are not using the full resolution of the sensor. You would have 10 pixels subtending the minimum resolvable feature of the objective, where only ~3 pixels would give all the information required for ideal sampling. In this scenario, and considering a Nyquist sampling rate of 3, I believe the resolution of the sensor would only be about 30% utilized. For a given sensor size, in MP, you have to choose a sensor with the proper pixel size. If you already have a well matched sensor, increasing the number of MP of the sensor will not make the image sharper. This is why it is important to match sensor pixel size to the magnification, or NA, of an objective. In the case of this camera, the pixel size is too small. A full frame, or APS-C sensor, with MP selected for proper sampling would be optimum.

Also, with images sized to the monitor, a 10MP image will not look any sharper than a 5MP image since computer monitors are limited to about 2MP, or less. So, how you are looking at the image is also important. Prints can sometimes be a better way to compare the resolution of two images. A resolution slide would be the best, but they are very expensive. The minor tics on mine are only 10um, so that doesn’t help much.

One way to tell if the sensor is running at full pixel resolution is to see if the refresh rate slows down when you increase resolution. At 10MP, the frame rate will only be around 1-2f/s, whereas it will speed up to perhaps 12f/s at 1.3MP resolution.

In answer Ron's question about pixel density of a Bayer filter/mask sensor, resolution is certainly lost. Since the array of red, green and blue is a 2X2 matrix, the linear resolution is reduced 0.5X for an unprocessed image. With some computational magic, a process called demosaicing can restore some of the lost resolution to the image. All modern cameras, include a image processor engine that performs a demosaicing routine on the image in real time (sort of), as well as other functions. The resolution achievable by this method is up to 0.8X the sensor resolution. So, if everything is done right you could get 8MP resolution from a 10MP sensor. Other problems are induced into the image by this process, such as Moire patterns, so it’s not without cost. There are also other tricks that some manufacturers use to increase the resolution by reducing the space between the pixels, such as making them into hexagonal tiles. These are more specialized than the Aptina chip that is used in the camera in question.

I haven’t been able to post lately, due to work load, but couldn’t help responding to yours, as this stuff is extremely interesting to me. I have a couple of these cameras sitting around and if I can find time, I will run a couple of resolution tests.
Cheers,
Gene
avatar Re: Amscope MU1000 camera - initial impressions
November 13, 2011 07:21PM
us    
Gene,

One other point, the resolution of the objective lens is not tied to N.A. but depends on the lens aberrations remaining in the design and manufacturing of the lens. Even the diffraction limit for a lens, which is the ultimate limit, is determined by the lens diameter rather than the f/ or N.A. Determination of the "pixel density" of an objective lens is not easy and could require some sophisticated testing.

Ron
avatar Re: Amscope MU1000 camera - initial impressions
November 14, 2011 08:39PM
us    
Ron,

Because there is some confusion regarding microscope resolution, I am going to compare both telescope and microscope resolution in order to put things into perspective.

Aberrations are present in optical systems and are a result of the design, materials and manufacturing process. They can always be reduced by better design and improved manufacturing. With modern day computer aided designs, state of the art glass and improved manufacturing technologies, high quality telescope and microscope objectives can be diffraction limited. In this case, the limit of resolution is not set by aberrations, but is limited by the immutable laws of physics, i.e. the wave nature of light (diffraction).

The angular resolution of a telescope is approximated by R = Lambda / D, where: R = angular resolution (arc sec), Lambda = wavelength of light, D = diameter of objective lens. Your comment that resolution is a function of D is correct for the case of a telescope objective.

However, we are interested in microscope objectives and that is somewhat different. Because the subject is not at a great distance or infinity, as with the case of the telescope, the expression for resolution must be derived for linear resolution. The result, in this case, is that the numerical aperture determines the resolving power of a microscope objective. The most common mathematical derivation for the resolution of a diffraction limited microscope objective is approximated by R = (0.61 X Lambda) / NA, where R = resolution in linear distance (nm), Lambda= wave length of light (nm) and NA = Numerical Aperture. The numerical aperture can also be shown to be: NA = n(Sin(A/2)), where: A is the angular aperture, n is the index of refraction of the medium through which the sample is being viewed. Note that you do not see the diameter, D, in the equation for resolution.

The resolution of a microscope is given in linear distance, whereas the resolution of a telescope is given in angular distance. The key here is to note that, for telescopes, the subject is at a large distance from the objective, usually at infinity. Therefore, the acceptance angle is very small, usually in arc seconds. For microscopes, the subject is very close and its distance from the objective is in the same order of magnitude as the diameter of the objective and dictates a large acceptance angle. The resolution is dependent upon both the distance to the subject and the diameter of the objective, not just the diameter of the objective.

You can look through microscope literature and microscope manufacturer’s data and see that numerical aperture is the parameter that they use to specify the resolving power of an objective. In fact, if you look at some microscope objectives and you will see that the diameters of many high resolution objectives are smaller than that of lower resolution objectives. This may seem paradoxical, but is easily explained in the derivation of NA. Finally, in the case of not so highly corrected objectives, many manufacturers lower the NA, so that it is still an accurate measure of resolution.

I am not sure as to what you are referring to as pixel density of an objective lens. Pixel density is a property of a sensor, not of a lens.

Gene
avatar Re: Amscope MU1000 camera - initial impressions
November 14, 2011 10:11PM
us    
Gene,

The resolving power of an objective is usually referred to the object plane and an equivelant "pixel density" can be computed. The camera pixel density will relate to this taking into account all of the intervening optics and the effective magnification. It is not a simple comparison.

In my opinion it is best to try to have a camera sensor with as high a pixel density as possible and then choose an objective an eyepiece to take advantange of the camera resolution. In addition, the linearity, distortions, and other factors must enter into the evaluation of a good camera - microscope match for good results.

Ron
avatar Re: Amscope MU1000 camera - initial impressions
November 15, 2011 02:26AM
us    
Ron,

This is getting to be too much fun, though time consuming. I hope that you are taking this in good spirit, as I certainly am. Hopefully, we can both learn something....

The resolving power of a microscope objective is given, by the manufacturer, for the objective alone. How can they know what you are going to do with it? They don't care what additional optics you are going to add to the system, and they certainly don't know what sensor you are going to use. So, the stated resolution of an objective is just that, the resolution of the objective alone.

Selecting a camera with the highest pixel density can be a mistake as I pointed out in my first post above. If the pixels are too small, then your image will be oversampled (Nyquist). The result will be that of a lower pixel density camera.

IMO, the way to set up a system is to first determine what you want your field of view to be. Then you have to decide upon the tradeoffs between sensor size and objective magnification to satisfy the field of view desired. Generally, the largest pixels that you can get, and still meet Nyquist criteria, will be the best. And, large pixels give the lowest noise. So, using one of the new Canon 120MP full frame sensors, for example, would produce grossly oversampled images and the result would not only be noisier but would not be any sharper than say a 21MP image. More pixels does not make better images.

A properly selected objective projecting directly onto the sensor produces the best images. You don't need intermediate optics (except with infinite focus objectives) or eyepieces. The finest micro photography produced today is by this method, using full frame sensors and objectives in the range of 4X to 40X. 40X is pushing things and requires a great deal of care to get good results. Of course, higher magnification objectives also means less depth of field, dictating larger multi-focus stacks.

Gene
avatar Re: Amscope MU1000 camera - initial impressions
November 16, 2011 05:31PM
us    
Gene,

The Nyquist criteria is, technically, the minimum samples required to accurately reproduce a sine wave of a particular frequency. No lens has a rectangular transfer function, so oversampling relative to the Nyquist criteria is not necessirally bad!

Your point of too high a sensor density is valid from an overkill, and noise standpoint, but down sampling with a good software algorithym can regain the S/N ration that a lower density sensor array would yeild.

The major reason to keep the sensor array as pixel density to a "reasonable" number is cost.

Ron
avatar Re: Amscope MU1000 camera - initial impressions
November 18, 2011 03:01AM
us    
Hello Ron,

Nyquist criteria is not correctly stated as “the minimum samples required to accurately reproduce a sine wave of a particular frequency”. The Nyquist theorem states that you can reproduce any complex waveform, in this case a function of spatial frequencies, by sampling it at a frequency that is at least two times the highest frequency contained in the band limited waveform. Band limited meaning that the Fourier series for the function must not have an infinite number of terms. Because physical systems have a finite number of terms, good reproducibility is usually achievable. The Nyquist - Shannon sampling theorems are far reaching and cannot be described in just a few words. The mathematical treatments are explicit and rigorous.

Let me try to simplify. We are talking about Nyquist sampling rate for the image plane, which in our case, is coincident with the sensor. That is, an image is on the sensor and we want to determine the optimum Nyquist sampling rate. For that reason, the optics and their transfer functions are not relevant. We just want to reconstruct the image on the sensor as faithfully as possible by choosing the best Nyquist sampling rate. The literature states that number is 2.5 – 3 for imaging sensors. I will agree that oversampling is not necessarily bad, only that is not an elegant solution, nor is it economical. For the case in question, the MU1000 camera (remember that is what this thread was originally about) pixels are wasted on a low resolution, NA~0.1, stereo microscope. A sensor, of the same physical size, with fewer and larger pixels would give better results and would be more economical. Binning (down sampling) is possible, but at added hardware or software expense. It can be a good thing to do if you are already stuck with too many pixels, but choosing the proper sensor to begin with makes more sense. And this is, in effect, what Henry is doing when he runs in a lower resolution mode of the 10MP camera.

This image is an example of the results that are achievable with just a 1.3MP sensor that has been optimally matched to the objective.

© Minresco


Since no one else has chimed into this thread, I can assume that no one is interested in our blathering, so it has become a waste of bandwidth. At that, I thank you for the stimulating exchange! Perhaps we can find more agreement, at another time and on a different subject. I suspect that Henry certainly didn’t expect this when he first posted his observations, regarding his new camera.

Cheers,
Gene
avatar Re: Amscope MU1000 camera - initial impressions
November 18, 2011 05:10PM
us    
The reference to band limited fourier decomposition of complex waves is a way of saying that there is an upper limit to the frequency of interest of a sine wave. In other words the Nyquist sampleing rate of 2 samples per cycle is appropirate for an accurate reproduction of that sine wave and the corresponding complex waveform that is being represented by its fourier spectrum.

We are rapidly deteriorating into a battle of "one upmanship" while saying the same thing. So I will call this exchange at an end and say "adios".

Ron
Re: Amscope MU1000 camera - initial impressions
November 19, 2011 03:38AM
us    
Gene,

Out of curiosity, can you post the details of the photo you used to illustrate your point? I used to take images like that with film and enlarging lens on bellows extensions. Have really never duplicated that with digital imagers.Of course, I'm usually trying to image some 0.1 mm speck!

Henry

Henry Barwood
Troy University
Troy, Alabama USA
avatar Re: Amscope MU1000 camera - initial impressions
November 22, 2011 08:56PM
Hello Henry,

I see this discussion with some interest, as I still wait to find affordable high resolution microscope camera. I would also like to make good pictures from a 0.1 mm speck. But with optical photography the limitation is the resolution of the lens. Practical the limit is in the 3-1 µm range depending on the N/A of the lens and its quality and it is more difficult to be vibration free in that range. So you will get at maximun 100 sharp data points for 100 µm. Well, some can be done with picture processing. So for me now the end for reflected light photography is in the field of view range of 1 mm, well sometimes 0.5 mm, but a still-picture with less than 1000 sharp pixels suffers from sharpness. (transmission microscopy is a different story)

I adapted a Panasonic Lumix D2 on a Leitz Ortholux I body (no relay lens), this works reasonable, and I believe that you can match this adaption only with a very expensive high end microscope camera. Interestingly the key was to minimize camera vibration. Typical I reduce the 4000 pixels of the camera in post process to around 1000.

I still hate that I have a shutter and wait for the day the vido grabber cameras go into the 12 Mpixel range.

Volker
Re: Amscope MU1000 camera - initial impressions
November 23, 2011 01:09AM
us    
Dear Volker,

Good points all. My best lenses were the old anastigmats. They suffered from some pincushion distortion, but had very high contrast. With a high saturation film like Kodachrome, I got beautiful images. Vibration was/is the bane of my existence also.

I would love to try one of the cooled CCD cameras (shutterless) such as are used for fluorescence microscopy and astrophotography, but my budget doesn't allow for this luxury.

Henry Barwood
Troy University
Troy, Alabama USA
avatar Re: Amscope MU1000 camera - initial impressions
November 23, 2011 02:52AM
Thank you all for this wonderful discussion. While it may have been "quiet" I am sure many, like I, were reading each post, and trying to digest the details.
Joe
avatar Re: Amscope MU1000 camera - initial impressions
November 24, 2011 02:26AM
us    
Hi all,

I used the previously posted image just to illustrate that even a 1.3MP camera can yield pretty good images if the camera sensor is a good match for the optical system being used. I’m not trying to imply that just matching a sensor to the optics is going to produce excellent images, but that this is one aspect of making a good image, especially if you are using a small size sensor.

In the case of the previously posted image, the microscope used was a Nikon SMZ800 with 1X ED Plan Objective (NA= 0.09), which gives a maximum resolution of ~3.7um on the object plane. Actually, I have measured it to be a bit better than the calculated value. The scope was set to 4X and it has a 0.7X relay lens in the photo tube, yielding a total magnification of ~2.8X. Referred to the image plane (the sensor), a feature that is at the limit of resolution (3.7um) would be 2.8 X 3.7um = 10.4 in extent. Since the pixel size for this camera sensor is 3.6um X3.6um, the Nyquist sampling rate would be 10.4/3.6um ~2.9. In other words, the image of a minimum discernable feature, referred to the sensor, covers 2.9 pixels. This satisfies the Nyquist minimum sampling rate, which is ~2.5 to 3.

From the above, it can be seen that if the pixel size is increased, the image becomes under sampled and resolution is lost. If the pixel size were reduced, the image becomes oversampled, and there is little or no gain in resolution. With the 10MP camera (1.4um pixels) the sampling rate would be ~7, which isn’t necessarily bad. However, other factors come into play with smaller pixels. Each of the smaller pixels has only ~0.15 the area of the 3.6um pixel. The sensitivity of the sensor is directly proportional to the area of a pixel, so the smaller pixel will require ~6X more light to achieve an equivalent exposure. Also, the well depth of the smaller pixel is less, so dynamic range suffers and signal to noise ratio suffers as well. Some of this could be corrected by clever hardware and/or software, but it will never be as good as that of the sensor with larger pixels.

There are a number of interacting parameters to play with when trying to optimize an image for proper sampling. These include the following:
Numerical Aperture (NA)
Magnification
Pixel size
Sensor size

Object illumination is related to the above, as well. For sensors with small pixels the signal to noise ratio is improved, up to a point, by using higher intensity illumination. Illuminators that are built into the microscope are practically useless. Halogen fiber optic illuminators are best and can be used with diffusers and still provide adequate illumination.

As further examples of what can be done with a small sensor, I have added two more images taken with a 1.3MP sensor. Keep in mind that a larger sensor, such as a full frame or APS-C, with larger pixels can certainly produce far better images. These larger sensors have up to 20+MPs, with pixel sizes up to 7.2um. Some of these sensors have areas that are up to 35 times greater than the 1.3MP sensor is being discussed here. But, for a camera that costs $129, the 1.3MP camera performs admirably.


© minresco
© minresco

Please forgive any mistakes in the above, as I had to fire this off quickly so as to get some shut eye.

Happy Thanksgiving!
Gene



Edited 1 time(s). Last edit at 01/05/2013 04:01AM by Mineralogical Research Company.
naykyar
Re: Amscope MU1000 camera - initial impressions
January 24, 2012 05:04AM
Hello,
I 'm naykyar from Myanmar.
I would like to request for Amscope MU1000 camera software, may i?
I bought Amscope T360B 10M with MU 1000 camera but havn't install CD.
Now i didn't use with Computer.sad smiley
Please help me.

smiling smiley
Attachments:
open | download - MU 1000.jpg (108.5 KB)
avatar Re: Amscope MU1000 camera - initial impressions
January 24, 2012 07:41PM
us    
Hello Naykyar,

You can download software, ToupView_EM, that should work with your camera from here.
If you want a simpler to use software, then down load the Light Version.

The first listing at the top of the page should be the correct version for your camera. Also, see the installation information at the bottom of the page.

Gene
naykyar
Re: Amscope MU1000 camera - initial impressions
January 27, 2012 04:41AM
Hello Gene ,

Thank you for your help.
Now i'm ok.
Author:

Subject:


Attachments:
  • Valid attachments: jpg, gif, png, pdf
  • No file can be larger than 1000 KB
  • 3 more file(s) can be attached to this message

Spam prevention:
Please, enter the code that you see below in the input field. This is for blocking bots that try to post this form automatically. If the code is hard to read, then just try to guess it right. If you enter the wrong code, a new image is created and you get another chance to enter it right.
CAPTCHA
Message:

Mineral and/or Locality  
Search Google  
Copyright © Jolyon Ralph and Ida Chau 1993-2014. Site Map. Locality, mineral & photograph data are the copyright of the individuals who submitted them. Site hosted & developed by Jolyon Ralph. Mindat.org is an online information resource dedicated to providing free mineralogical information to all. Mindat relies on the contributions of thousands of members and supporters. Mindat does not offer minerals for sale. If you would like to add information to improve the quality of our database, then click here to register.
Current server date and time: July 30, 2014 01:02:41
Mineral and Locality Search
Mineral:
and/or Locality:
Options
Fade toolbar when not in focusFix toolbar to bottom of page
Hide Social Media Links
Slideshow frame delay seconds