maoyaowu 发表于 2013-4-20 15:07

【译文】业余天文望远镜的平场文件与漫射光

本帖最后由 maoyaowu 于 2013-4-20 15:56 编辑

基于这篇平场相关的文章,总结如下:
1.      望远镜的接环,附件等消光要做好。2.      LRGB每个通道的平场文件使用各自通道的滤镜拍摄。3.       IR只是漫射光中的一部分。4.      月光明亮的夜晚也可以进行深空拍摄(是么!? 没有试验过拍摄效果,有经验的朋友们讲讲)。
5.       发现CCD上有些小瑕疵(如芯片金属焊接线反光等)要淡定,不要认为是相机有问题。6.       不要过于苛求,尽力做到最好就行。



Flat Fields andStray Light in Amateur Telescopes业余天文望远镜的平场文件与漫射光March 8, 2013:12:00 AMFlatFields and Stray Light in Amateur Telescopes
(The Ugly Truth - Its more Complicated than you ever thought!)业余天文望远镜的平场文件与漫射光(丑陋的真相-----比你想象的更复杂!)
Alan Holmes
3/8/2013A lot of users struggle with flat fields thatdon’t work well, and leave gradients in their images, or hot spots, or otherhard-to-process-out artifacts. As a result they have resorted to twilight flatsand other techniques to get better results. In this paper I will reveal thereal source of the problems, discuss why twilight flats are not very good, andoutline some tips for better results. I believe even those users who think theyhave mastered the flat-fielding technique still have problems they do notrecognize.
许多用户一直在纠结平场文件的问题:平场文件叠加到亮场文件后效果不好,在亮场文件上留下了亮度变化,热点,或者其他难以去除的人为瑕疵。基于此,他们已经重新开始使用天文晨光或其他技术去拍摄平场文件以取得更好的结果。在这篇文章里,我会揭示该问题的真正根源,讨论为什么天文晨光平场不是很好,并且罗列一些获得更好效果的小窍门。我相信就算哪些认为自己已经掌握了平场技术的用户们也会存在一些他们不愿意承认的平场问题。
To begin – many black surfaces are not
I have constructed a flat field system at homewhere I can use LEDS of different colors to take a really good flat fieldwith our CCD cameras, and telescopes. It is nothing more than multiple LEDS ofdifferent wavelengths, and a large pasteboard box painted black on the inside,with apertures sized to recreate an F/6 system. In Figure One I show flatfields collected using an expensive refractor any of you would be proud to own,using green light and near IR (940 nm) light. There is no vignetting whateverin this system! What looks like vignetting at the edges is actually excesslight in the center of the image that has glinted off the coupling tubes nearthe camera. The donut appearance in the IR is due to these same glints.
开始—许多黑色的表面并不足够黑
我在自己的家里建造了一个平场拍摄系统,以便能够使用不同颜色的LED光源拍出真正很好的平场文件(使用天文望远镜,和CCD相机)。拍摄器材包括:不同光波长的LED灯,内部涂成黑色的大的亮场盒(制作均匀光源的盒子),F6焦比的望远镜。图一的平场文件是用一个大家拥有都觉得很牛X的昂贵的折射镜拍摄的,使用近绿光和IR(940nm)的光。这个系统拍摄的平场完全没有任何渐晕变化!左边图像边缘看上去像是渐晕变化的部分其实是中部过多的光被与相机连接的转接环反射的结果。右边图像圆环状的部分也是同样的转接环反射的结果。
FigureOne: Flat Fields with an Expensive Triplet Refractor图一:昂贵的3片3分离APO折射镜拍的平场


The center to edge variation here is about 6%,which would require flat fielding in an astro-image. The problem is lightpollution for most people means that their sky at night is a greenish yellowcolor, but if they took a flat field using a flashlight with an incandescentbulb, which has strong IR, the astro-images would have an inverse donutillumination pattern. There is another concern here. Many astrophotographersare doing photometry with their systems, and quite often compare two stars indifferent parts of the field to make their measurement (differentialphotometry). They would flat field their data assuming the pattern seen in thegreen image was vignetting, which means a sensitivity dropoff at the edges, butit is not. Sensitivity is actually constant to the edge. Figure Two illustratesthis problem.
图片中,中央到边缘的亮度变化大约是6%,这在天文照片上是需要叠加平场文件了。问题是,对于大多数人来说,光污染意味着,他们的夜空是黄偏绿的颜色,如果用白炽灯做光源拍摄的平场,因为白炽灯含有强烈的IR光线,天文照片会有反向的圆环状特性。还有另外一个问题。许多天文摄影人使用他们的系统做光度测量,并且经常需要对比照片中不同部分的两颗星来进行测量工作(示差光度测量)。他们会对照片做平场处理,因为他们认为绿色的照片肯定是渐晕的,在照片的边缘会有微弱的减光,但实际上不是这样。边缘的感光度跟中心部分是一样的。图二演示了这个问题。
FigureTwo: Stray Light looks Like Vignetting图二:漫射光看上去有渐晕



This actually occurs with photometry. To properlymap out the sensitivity of the image one must move a star around the field andmeasure its brightness at many places. Or, eliminate the stray light!
这是在做光度测量时的实际情况。为了适当的映射照片的感光度,必须把一个星点在照片不同的地方做测量,或者,消除漫射光。
The stray light, particularly in the infrared, isusually caused by the manufacturer using black anodized aluminum surfaces inthe telescope. Black anodized surface have about 50% reflectivity at 940nm. Figure Three shows a number of "black" objects at 940 nm, as wellas a white business card. You "see" the problem, literally. Trulyblack surfaces would be as dark as the interior of the nosepiece on the rightside of the image.
漫射光,尤其是近红外部分的,通常是因为设备制造商使用黑色阳极氧化铝做表面。黑色氧化铝表面对IR(940nm)部分仍然有50%的反光。图三展示了一些“黑”色的配件,还有一张白色的名片。你看,问题在于,真正的黑色,应该像右边的接环中间部分的黑色一样黑。
FigureThree: "Black" Objects at 940 nm图三: “黑色”物体IR(940nm)


It is actually easy to see if you have someversion of this problem. Take the camera out of the focuser position, point thecamera at the sky or white surface, and look into the tube. Put your eyewhere the camera was. What you should see is the telescope aperture floating ina dark void. With the high end refractor of Figure One, I saw the patternillustrated in Figure Four.
很容易检查你是否存在这些问题。使相机对着天空或白色表面,略微失焦,然后取下相机,观察镜筒。你看到的是,望远镜的光圈被黑色的虚空包围。我使用的是拍图一时使用的豪华高端APO,我看到的就像图四所示。
FigureFour: Looking up the Tube图四:观察镜筒



As you can see, I had lots of glints off thecoupling rings. I made a special camera for taking this image by modifying anST-3200 to have a 6 mm F/12 focal length pinhole lens right in front of it. Itwas revealing, once again literally. Below I show the results for a ChineseNewtonian telescope. You can see the aperture and the secondary spider in themiddle. However, that glow around it is 13% of the total light hitting thefocal plane at 940 nm.
就像你看到的,有很多双环亮光圈,我使用了其他的相机和牛反望远镜也做了拍摄,效果如图五
FigureFive: Chinese Newtonian图五:牛反效果


I know many of you are thinking "Why do I needto worry about this – the sky is dark?" The answer is, it’s not. You arepointing your telescope into a hemisphere of light. It may be dim, but so iswhat you are imaging. In fact, for the high end refractor, the fraction of theskylight that you want to hit the chip is only 1/7300th of the light enteringthe lens, and you want to knock down that unwanted stray light to the 1% level,or almost a million times total.
我知道你们许多人会想“为什么我要担心这个问题-天空是黑色的?”回答是:不是的。你的望远镜指向的是一个光半球。天空是暗谈的,这就是你拍的东西。事实上,就算是高端的折射镜,透过望远镜镜片的光线,只有1/7300投射到了CCD芯片上,你肯定想把不需要的漫射光减少到1%的级别,或者几乎十千万分之一。
Now it is appropriate to delve into the sciencebehind this problem. The stray light problem would actually flat field out fineIF the flat field light source were exactly the same spectrum as the night sky.However, this is not the case. In Figure Six I show the relative spectrum ofseveral light sources, measured with an SBIG DSS-7. My night sky from my backyard is also included.
现在,是时候深入研究这个问题背后的科学了。漫射光的问题真正会导致平场效果不是那么的完美,如果平场光源的的光谱跟夜空的光谱相同然而这还不是问题原因的全部。图六,我展示了不同光源的光谱。光谱是使用SBIG的DSS-7测试的。我自家后院的夜空光谱也包括在内。
FigureSix: Spectrum of Typical Light Sources图六:典型光源的光谱分析



A wavelength of 4500 Angstroms (450 nm) is deepblue, a wavelength of 5500 Angstroms is green, and 6500 Angstroms deep red.However, CCD cameras have sensitivity up to 10000 Angstroms (1000 nm), and thesky has about half its brightness, for a dark site, above 7000 Angstroms. Infact, the older imagers among you will remember the early days of CCD imagingwhen users marveled at the fact that moonlight was not a big problem forimaging. That was partly because it was a small fraction of the total skylight,which included the infrared. By the way, I have also measured the twilightspectrum, and it is a deep blue weighted spectrum, as one might suspect. It isa poor match to the night sky. Also, I have had some users that were flatfielding each color by itself to try to beat the flat field problem, and usingincandescent bulbs turned down low. They were taking blue flat fields withorange light! That would work except most filters have some IR leakage, and whenthe IR content of the light is 100 times the blue, the leakage is as big as thedesired light.
波长450nm是深蓝色,550nm是绿色,650nm是深红色。然而,CCD相机对上到1000nm的光线都敏感,在一个黑暗的地点,天空光线的波长可以到达700nm,有亮度的一半。实际上,资深的天文拍摄者会回想起早期的CCD天文摄影,他们惊奇的发现月光并不是天文摄影的一个大问题。这是因为月光只是总天光(包括红外部分)中的一小部分。顺便,我测试了一下天文晨光的光谱,是偏重深蓝色的光谱。跟夜空的光谱相去甚远。我也发现一些用户是用降低亮度的白炽灯做光源对每个不同的色彩通道拍摄平场文件以避免以上的问题。他们用橘红色的光源去拍摄蓝色通道的平场文件。在滤镜没有IR泄露的情况下,可以工作的不错,当光线中IR部分超过蓝色光线100倍以上的时候,泄露的IR光线就跟你需要的颜色的光通量一样了。
The other important part of this issue is thatdeep sky objects all have very low contrast against the sky background. FigureSeven shows a picture of the Horsehead Nebula captured by Tony Hallas with aluminance filter on a dark night from a good sight. Note that the interestingfeatures require detection of objects at the 1% contrast level, and world classwork like Tony’s reveals features down to 0.2% and less. By the way, thebrightness of areas of an astro-image are easily measured using CCDOPSCrosshair mode.
该问题另一个重要的部分是,深空目标与天空背景的对比度很小。图七展示了托尼.哈拉斯使用L滤镜在一个很好的拍色地点拍摄的马头星云。请注意:图片中如果有能引起兴趣的东西,需要至少1%的亮度差异,托尼作为世界级的高手可以做到0.2%甚至更低。图片中区域的亮度可以用CCDOPS的十字模式轻易的测出。
FigureSeven: Tony Hallas Horsehead, Luminance Filter图七:托尼.哈拉斯用L滤镜拍的马头星云



TheSolution – Make Those Surfaces Black!How to solve this problem – paint your interiorsurfaces! I have measured one paint that seems to work well, Rust-oleumSpecialty High Heat Barbeque Flat Black spray paint. Others are no doubt good,but this one I know is flat, and has about 10% reflectivity in the IR. Avoidgarden variety black paints since they may contain a dye, rather than apigment. Many dyes common in the printing industry and elsewhere turn white at700 nm. See the black felt in my previous image for proof. The Barbeque Blackis good since dyes can’t take the heat, and the formulation uses a pigment.Remove the interior surfaces from any optical elements, and paint them well,masking the threads with tape. As an example of how well this can work, seeFigure Eight below. The user was getting the flat fields shown on the right,and was convinced he had a defective camera, and wanted a new STL-11K. Aftermany discussions, he painted the interior of his new focuser (a good brand, bythe way) with flat black paint, and obtained the flat field on the right –problem solved!
解决方案—想办法把相关配件的表面弄的更黑怎么解决这个问题呢-把你的光学系统的内部表面涂的更黑些!我测试过一种涂料看上去不错,Rust-oleum特制的烤黑色高热喷射涂料。其他的也应该很好,但这个涂抹的很平坦,对IR的反射只有10%。避免了其他各种涂料中包含的是染料而不是颜料(含有较大的矿物质颗粒)。许多目前在印刷工业中常用的染料在700nm光线下看起来会变白。以前的图片显示的结果就是证据。烤黑色工作的很好是因为染料不能吸收热量,烤黑色是使用的颜料配方。用这种牌子的涂料把配件内表面仔细喷一下,记得喷的时候把配件的螺纹部分包好。比较一下图八,左边是没有涂黑配件内表面时候拍摄的平场,当时他都认为自己应该把去整一个新的STL-11K。右边是涂黑配件内表面后拍摄的平场—问题解决了。
FigureEight: Case History – Before and After Painting Interior of Focuser图八:电动动调焦内部涂黑前后拍摄的平场文件对比


It is also a good idea to use a light source forflat fields that is somewhat similar in spectral shape to the night sky fromwhere you live. We hope that some company will take it upon themselves to makea hyperspectral flat field source using an array of LEDs with different centerwavelengths to accurately reproduce the night sky spectral shape. SBIG has theequipment to measure the spectrum of the night sky or various light sources.Our DSS-7 has existed for many years, and our new ST-I Prism Spectrograph willbe a valuable accessory enabling such measurements.
使用与你拍摄地点的夜空的光谱形状相似的光源拍摄平场文件也是一个好主意。我希望有些公司能够使用一组不同中心波长的LEDS灯生产可以准确模拟夜空光谱。有些CCD相机具备测试光线光谱的能力。略。。。。。。
It’s Not Always Just Paint!
I may have led the user to believe that with asimple can of spray paint his problems are solved. As with most things in life,it is not quite that simple. I have made some effort to determine the limits ofCCD cameras to detect low contrast objects, and have found that at the 1% levelthe CCD itself has spectral variations of sensitivity across its surface. NoteFigure Nine, revealing bond wire glints. Light reflects off the gold structureand bond wires at the edge of the CCD, then reflects off the CCD cover glass,and produces this effect. Figure Ten shows a Single Shot color ST-8300 image ofa galaxy from a light polluted sight with the contrast pushed, showing furtherevidence of the metallization changing the color of the sky background near theupper and lower images of the image. All of you reading this paper have thisproblem – check your images.
并不总是仅仅涂黑的问题!
我可能已经误导用户认为用一小罐涂料就能解决他们的问题。就像我们日常生活中遇到的其他问题一样,事情远远不是这么简单。我曾经努力去确定CCD相机在探测低对比度物体的能力极限,发现在1%对比度的级别上,CCD相机的芯片表面存在不同地方光谱敏感度不同的情况。注意图九,揭示芯片结合线(半导体封装时的焊线)导致的不正常亮度变化。芯片边缘的芯片焊线轻微的反射了CCD玻璃表面的光,形成了这种效果。图十显示了ST-8300C在一个有光害污染拍摄的星云图片,进一步显示了芯片金属焊线造成的天空背景颜色变化,在图片的上、下部有明显的痕迹。所有读这篇文章的人都存在这个问题-检查你们的照片。
FigureNine: Bond Wire Glints图九:半导体封装时的焊线的微弱反光



FigureTen: Single Shot Color ST-8300 Image under Light Polluted Conditions图十:ST-8300C在光污染情况的照片


Sometimes the CCD itself has issues. Many of ourusers are pushing the limits, taking images from places like Tokyo or Beijing,where the contrast of deep sky objects against the night sky is 50 times poorerthan what Tony Hallas deals with. An example of what can happen is shown inFigure Eleven below, a green STL-11K flat on the left, and a red STL-11K flaton the right. Note the cold spot in the center of the CCD in the green imagebecoming a hot spot in the red. A problem like this is hopeless to flat field,even with a hyperspectral flat field source. Unfortunately this chip is notconsidered defective by Kodak (now TrueSense). At the time of this user’sproblem I surveyed a number of STL-11K cameras from our production line anddetermined that this was truly an unusual camera in the magnitude of thiseffect. However, all of them had variations at the 1% level, but not soprominently placed as the center of the CCD!
有时候是CCD自身存在问题。许多用户在不断的接近极限,他们在东京或者北京拍摄,哪里的深空目标与天空背景的对比度要比托尼.哈拉斯应付的恶劣50倍。图十一显示了会发生什么,一张STL-11K拍摄的绿色通道平场,和右边一张STL-11K拍摄的红色通道的平场。注意绿色通道平场中心部分的的冷点,在红色通道平场中变成了热点。像这样的问题在拍摄平场的时候是没有办法回避的,哪怕是使用超级理想的平场光源。很不幸,有这样问题的芯片不会被柯达(现在是TrueSense)认为是有瑕疵的。发现这个问题后,我调查 了产品线上的几台STL-11K,发现,对于这种程度的效果来说这确实是一个不常见的相机。然而,这些问题在程度上有1%的变化,位于CCD的中央也不是那么的引人注目。
FigureEleven: STL-11K Flat Field with Unusual Problem图十一:STL-11K平场存在的不常见问题



Most amateurs take color images using the LRGBtechnique, where the L (luminance) image is painted with the color from the RGBimages. There is some debate about whether one should use a clear filter, whichpasses all wavelengths, or a luminance filter, which blocks the infrared. Notethe flat fields below in Figure Twelve, captured using the popular KAF-8300CCD. A green flat is on the left, and looks quite smooth except for a dustglob. The flat field on the right is with an IR LED. Note the polishing markson the right, which are revealed by the greater penetration of IR light intothe silicon, the increased gradient left to right, and the grid structure becomingmore prominent. Both images are displayed at the SAME 10% contrast full scale.This is a good chip!
大多数的业余爱好者使用LRGB方式拍摄彩色照片。一直存在一个讨论:是否应该用放通所有波长光线的C滤镜,或者用带红外截止功能的L滤镜来拍摄L图像。注意图十二,下面的平场使用流行的KAF-8300CCD拍摄。绿色通道的平场在左边,除了几处灰尘痕迹,看上去很平整。右边的平场是用较多IR 光的LED做光源拍摄的。注意右边光滑的亮部,这是较多的IR光线投射到了芯片上,这增加了整张照片从左到右的的亮度变化,照片被明暗分隔成网格状的情况更明显了。两张照片都是全尺寸的10%。芯片没有问题!
FigureTwelve: ST-8300 Flat Fields图十二:ST-8300平场文件


By now you may be starting to feel CCD imaging ishopeless. That is far from the case. This last example merely illustrates thelimitation of broad band imaging, and points out the importance of trying tomatch the flat field illumination to the night sky spectrum. For now I wouldrecommend users take their "L" shots with a luminance filter to getmore easily processed results. We here at SBIG are committed to improving whatusers can achieve with their equipment, and it is a continuous process. Tokyoor New York may be where you live! However, the more daylight we can shed onthese mysteries of CCD imaging the more quickly solutions will be developed. Ishould note, though, that telescope suppliers need to pay more attention to bafflingand flat interior coatings in their products. Many are good, but there are somevery bad scopes out there. I will not name names in this paper. However, we mayoffer a short focus ST-I lens accessory soon that lets users make their ownmeasurements.
到现在,你可能感觉CCD天文摄影是多么的不可救药。这还远不是问题的全部。最后一个例子仅仅演示了CCD宽带(LRGB)摄影的局限性,指出平场拍摄光源的光谱特性与夜空的光谱特性一致的重要性。现在我建议用户使用L滤镜拍摄L通道而不是C滤镜。因为这样拍摄的照片更容易后期处理。略。。。。。。

a125278748 发表于 2013-4-20 15:29

太专业了,楼主转发辛苦

maoyaowu 发表于 2013-4-20 15:48

a125278748 发表于 2013-4-20 15:29 static/image/common/back.gif
太专业了,楼主转发辛苦

偶自己翻译的{:soso_e120:}

山脚落 发表于 2013-4-20 16:43

很好的资料,多谢楼主的贡献~

boneking 发表于 2013-5-20 23:31

楼主辛苦,感谢分享。
页: [1]
查看完整版本: 【译文】业余天文望远镜的平场文件与漫射光