Crossing datas from four sites helps to choose the Best Cameras :
BDP : Brendan Davey Photography.com DPR : Digital Photography Review.com Dxo : DxoMark Sensor Test Database.com P2P : PhotonsToPhotos.net Bill Claff
APSC DSLRs - with 4/3 or 1" stills Crop Modes ?
For Astro, Birds Photography, there are not so many APSC cameras available with a Crop Mode down to 4:3 or 1" format for stills. Now that they all get Fully Addressable CIS, CMos Image Sensor, RoI - Region of Interest - is a totally CostLess Option, but very useful for AstroPhotography of Deep Sky Small Objects or Birds photography !
Let's examine 2 pairs of APSC DSLRs through P2P.net, a Used and a New one :
Click Image, Ctrl+ to Enlarge
Used APSC Cameras : Pentax K-5(II(s)) betters the Eos 60d in every characteristic, except for the Read Noise where the 60d's slightly better. Canon DSLRs were Champs of the Conversion Noise : Eos-60d = 8.5 vs PK-5 = 0.6 ! Red Layer especially suffers from that Huge Conversion Noise level : a Stock Pentax K-5 with an UHCS filter shows www.dpreview.com/challenges/Entry.aspx?ID=1195990 at least equivalent Red Performance than an Expensive Eos-60dA, the proofs here : www.dpreview.com/galleries/7467909648/photos/4208343/pentax-k5-rosette Note that the 52mm secondary mirror of my 200/1000 SkyWatcher is way too narrow ! Pentax K-5's body is Weather Sealed, its specified min.operating temperature is -10°C. The Pentax K-5 also has a very low Dark Current level so that you can Expose until 30 minutes at ISO 80 by 12-25°C ambient Temperature and get so Clean Raw files that the Bias Maps, Dark Frames are useless and, in most cases, you'll only need 3-8 Raw files: www.DPReview.com/Galleries/Tags/ISO-80 17 min. Exposure through UHCS filter to darken the Galaxy, then pushed 6 Ev : www.cloudynights.com/gallery/sizes/32054-pk5lv-5-6/large/ Not a bit of Banding, nor any Fixed Pattern Noise. The Pentax K5 is IsoLess, far better results are achieved by stacking ISO 80 subs to get a High SNR picture that you can easily Push a lot than shooting at ISO 1600 as you have to with a Canon camera very Noisy at base ISO. 1st exceptions the last Eos R5 and R6 ! the imx071 exmor sensor inside the PK5 is so fine that chinese camera assemblers did not hesitate launching their Asi071, Qhy168C and so on, at the high price of $3000€ = 3 times, 3x the Price I bought my Pentax K-5 New in Spring 2011 ! Some models are still available at 1/3~1/2 of launch price, see how low is their imx-071's Dark Current level:
+ replace 'Cool' by 'Pro', add '+Manual' in Browser's Search-bar for more infos.
- first CMos "Exmor" sensor: ~15 years on the market and still In the Race.
The very 1st sensor w/o any Amp-Glow.
Click Image, Ctrl+ to Enlarge
New APSc Camera: if I had to buy an APSc camera for Long Exposure in Low Light or Astro-Photo today, I would certainly choose a Nikon D7500 over a Canon Eos 80d - the Eos 90d has not yet been tested by DxoMark. Pentax K3 I, II, III are Not at the DR level of the K5 ; they do Need Hard Raw Smoothing to follow Modern Dual-Gain BSIs. Proof, the K3-3 Mono w/ imx571 doesn't need that Strong Raw NR to gain 2+ Ev DR. The Eos 80d, now very close to the Pentax K5 is a high progress over the Eos 60d(a). The Nikon D7500 beats them all in all characteristics except for Long Exposure Noise. link to : Brendan Davey's Long Exposure Sensor Testings The Pentax K5 is still the Long Exposure King, far less Noisy than either the D7500 and the D500 which share ~very similar sensors, if not exactly the same. Canon Eos 80d is roughly in between both for long exposures, but way more noisy than both the Nikon and the Pentax for Exposures of 30 seconds or shorter.
Click Image, Ctrl+ to Enlarge
Now DxoMark don't allow comparing more than 3 camera. But that wasn't too difficult for me to add the Canon Eos 60d over the screen shots of the 3 APSc DSLRs, Canon Eos 80d, Pentax K5 and Nikon D7500, here sorted by their IQ performance rate.
Incredible, the Pentax K5 beats all these Flagships in almost every criteria measured by DxoMark ! Exception, the Nikon D3x is better of 1 bit Color-Depth in Portrait Mode, but, was launched at $8800 in the UK - more than 8x the price I paid for my new K5 ! The better ISO score of the Full-Frames is logical, it's because their 2,33x sensor Area, all other things being equal, the Dxo-ISO score is function of the sensor's 2 dimensions : Low-Light Photography and Sensor Sensitivity Comparison Tools for Astronomy You see the 24x36 are far from reaching ISO 2707 ! The Sonys are only 1,2x the K5 ! Dynamic Range: the ability to show NoiseLess Details, even in the Deepest Shadows, and here, the Pentax K5 shines : 2,4 Ev better than the 2012 Eos 5D III, 1,8 Ev better than the Sony Alpha-900, first camera to implement an On-Sensor ADC, an Analog to Digital Converter which prevents aSaP the Signal to be Polluted by Conversion-Noise : External ADC, Weak Filtering Power Supply, Weak Shielding ruin the Dynamic Range
Comments from DxoMark in their Review of the Pentax K-5 : « The best APS-C in all tested fields »
~~~~~ tbc... ~~~~~
FF (mf) cameras w/ APSC (ff) stills Crop Mode ?
For Astro, Birds Photography, there are not so many MF cameras available with Crop Modes down to FF + APSC formats for stills. Now that they all get Fully Addressable CIS, CMos Image Sensor, RoI - Region of Interest - is a totally CostLess Option, but very useful for AstroPhotography of Deep Sky Small Objects or Birds Photography !
Let's examine 4 FF 24x36 DSLRs at P2P.net, the reference being my old Nikon D600 :
Click Image, Ctrl+ to Enlarge
Comparing these 4 Full-Frame DSLRs, you see the old Nikon D600 could still be a winner. 14+ Ev Dynamic Range, i.e. 76+ e- Full Well Capacity vs Low Noise levels, especially the Conversion noise level which is as low as the value of the Nikon D850. DR is 2~2.5 Ev better than 2012 Eos DSLRs, D600 Low Light's ~664 ISO better : Canon-Eos-5D-Mark-III-vs-Nikon-D600-vs-Canon-Eos-6D-Perfs-Compared If you are buying a D600, a D610 just spend a little more for a D750, it will beat the D600, D610 in the Long Exposure Sensor Testing of Brendan Davey Photography : path to : Brendan Davey's Long Exposure Sensor Testings You may also enjoy shooting with this D750 at a High ISO setting - but, don't forget the imx-128 isoless Sony Exmor sensor is at its Best between ISO-100 and -400 !
Asi128MC-Pro & Qhy128C-Pro Astro cams launched at ~$6000€ are easy to find discounted or used for ~35-60% that price with imx128 really stunning performances
+ replace 'Cool' by 'Pro', add '+Manual' in Browser's Search-bar for more infos.
I find those Best FF Cameras don't show that much improvements over my 9 years old Nikon D600. So I put my wishes in a " Blue Dream Camera " at the right of the 5 DxO rated 100 from 3 different brands, with my old Nikon D600 as a Reference at left :
Click Image, Ctrl+ to Enlarge
Note that if the D600 has an overall score of only 94, this is mainly because of its lower Pixel Count, 24 versus 45 for these 5 recent FF cams. Plus 24 MP gives only 10 MP in APSC crop mode, this is Not enough at the time where Recent APSC cameras get 20, 24, 26, 28 MPix. D600's Color Depth is only 25 bit vs recent FF cameras are over 26, but despite its lower Pixel Count, D600 is still " in the race " for its Very Good Dynamic Range, especially at ISO 200, and Good Low-Light-ISO score.
How about the Signal to Noise Ratio, SNR, of these 5 Top Ranked FF cameras :
Click Image, Ctrl+ to Enlarge
SNR is one of the main criteria for Long Exposures in Low-Light, Astro-Photography. If you plan to Print Large Posters, compare camera's Screen Modes, vs if you expect the benefits from DownSampling, then compare cameras in Print Mode. Blue Curve show my expectations for a FF camera with More than 48 dB SNR at ISO 100.
How about the Dynamic Range score of these 5 Top FF cameras w/ a rating of 100 :
Click Image, Ctrl+ to Enlarge
Due to their very Low Noise levels, CMos Image Sensors, CIS cameras show much improved Dynamic Range vs older, mainly APS-C CCD cameras with Lots of Noise. Dynamic Range: the ability to show Noise-Less Details even in the Deeper Shadows. If you plan to Print Large Posters, compare camera's Screen Modes, vs if you expect the benefits from DownSampling, then compare cameras in Print Mode. The Blue dots show my expectations for a FF camera with more than 15 Ev of DR at ISO 100.
How about the Color Sensitivity score of these 5 Top Ranked FF cameras :
Click Image, Ctrl+ to Enlarge
For Long Exposure in Low-Light, Astro-Photography, Color Sensitivity is not the most important characteristic to take into account when choosing a new camera, nevertheless if you are also interested in shooting Architecture, Landscapes, Portraits, etc with it, you have to take this parameter seriously too. The Blue Curve shows my expectations for a Full Frame camera with More than 27 bit of Color Sensitivity at ISO 100.
Signal to Noise Ratio higher than 48 dB, Dynamic Range better than 15 Ev, Color Sen- sitivity higher than 27 bit, Low-Light ISO score better than 5680 ! Are my expectations, too high ? I don't think so, if you aren't ambitious enough to exceed your limits and usual limits, don't bother, buy any camera, else always buy the best tool for your requirements.
Now, I just have to keep waiting for a camera that'll reach these IQ performance levels...
A good way to evaluate the DxO ISO value of an unknown camera sensor, is to apply a rule of 3. Let's compute an ISO value for the Gfx100s' imx411 CMos, the closest sensor by Pixel Design recently tested by DxoMark is the Pixii A1571's imx571. Its ISO sports score is 1939. I will only keep 85%=1648 due to the fact a 4x bigger sensor has to fight more against Heat, Amp Glow, etc. The calculation from the dimensions : 1648 / 23.5 * 44 / 15.7 * 33mm = 6486 for the Fuji Gfx100s' sensor. The same rule of 3 I used on 16 April 2014 to evaluate the DxO ISO value of the Pentax 645z. I was very close : "4500" instead of 4505 published by DxoMark 3.5 years later. I was optimistic for the Dynamic Range, 15 ev instead of 14.7 ev and dead on for the Color Depth, 26 bit exactly ! Here are my Forecasts after the Pentax 645z announcement published by DPReview Pentax 645z : my sensor rating Forecasts, April 16 2014
Actually testing the Fuji GFX 100s, seems to meet my expectations 5/5 IQ wise. Here are the results of my first night testings during Christmas, New Year's day: Fujifilm GFX 100s, the first 7 night-tests edited
Shooting, everything was shot Handheld w/ GFX-100S and GF 32-64 f/4 JPGs. ISO Sensitivity : ISO 800-1000 = all in HCG mode range. Aperture : f/5,6-6,4 - SS : 1/10, 1/13-1/15 sec. thanks to the IBIS system.
Building, Stock Exchange, Wheel : perspective straightened in Elements 3. Editing, Size reduct°, 100% Crop insert°, USM, Save to JPG 96% : IrfanView.
Gallery Marseille by Night, shot with the Fujifilm GFX 100s Maximum 3000*2250 pixels or 2,5 MB cause ISP Speed's way Too Slow here, Dpr Time Stamps 80 min: Retries, Timouts, to Upload 6 JPGs, 12 MB ! So, a 100% Crop from the Fuji .JPG file's inserted inside each.
Datas about Cameras' Low-Contrast Resolution Ranking from www.chip.de : Low-Contrast Resolution (Line-Pairs /Image-Height) is very close to
Real-Life Results, opposed to ~Optimistic High Contrast Resolution values~. Then I put them in a sortable per column table in order you can
easily compare cameras. The table also shows interesting datas about Screen / Printed Visual Noise. It's readable here - MecAstroNics.free.fr/LCResol-ScnPrtVN-FFMlRx.html
* Calculation Method * datas from chip.de + "Horizontale Tabelle Anzeigen" - LCresXXX = Auflösung bei ISO-XXX * Detailtreue bei ISO-XXX / 100.
- Auflösung bei ISO-XXX = Max Resolution (in Line-Pairs per Image-Height) measured at ISO-XXX setting according to ISO-12233.
- Detailtreue bei ISO-XXX = The level of detail is a measure of how well fine details are retained in a subject with Low-Contrast ratio.
It sets the Low-Contrast Resolution measured according to ' Dead Leaves ' method (in Line-Pairs per Image-Height) in relation to the number of Line-Pairs of the ' Normal ' Resolution measurement at the same ISO-XXX level.
I'll build the same table from chip.de datas for APSC Cameras as soon as possible.
~~~~~ tbc... ~~~~~
Camera's other Features to take into account
Is IBIS a Must-have Feature or... a Penalty ?
I am always very surprised to see so many photographers, even calling themself expert, taking cityscapes at ISO 12800 and not seeing that in this way they lose more than half of the sensor performance of their camera, often an Expensive High-End model ! High ISO settings should of course be reserved for restless subjects, childrens, sports, etc...
If your camera does not include this IBIS feature, you are doomed to Zen still postures to get sharp pics w/o tripod at very low speed Hand Held, depending on the lens used.
Sometimes it's successful, 2nd photo : Opera de Marseille shot from a giant flower pot containing a palm tree, where I am climbed to avoid a row of parked cars in the frame, Nikon D600 + Samyang 35mm F/1.4 Hand Held = 1/25 sec. at F/2.8, at ISO 200, In depth Test, geo444 compares Samyang 35/1.4 - Nikon f2 - Sigma f1.4
Sometines you have no choice : Marseille, Chistmas 1984, my Bronica is loaded with ISO 100 Fujichrome slide film and... No Way to install the tripod in that Crossroads : Bronica GS1 +Zenzanon PG 100f/3.5 Hand Held : 1/8 sec. F/4 - ISO 100 www.dpreview.com/challenges/Entry.aspx?ID=972744
Next...
The classic Big DSLR or the tiniest Mirrorless Camera ?
But, some reduced size ML cameras, stuffed with electronic cards and which sensor is filled with Transistors to achieve On Sensor-PDAFC, PDAFS, EFCS, ES, NR, ADC, Big Buffer, High Speed Transfer, etc... Plus the IBIS system that Thermally isolates the sensor from the body, hence reduces the Heat Transmission from the electronics to the outside via the camera body... All these drive to Over-Heating issues during Video, or Self-Heating problems during Astro-Photography and/or Intervallometer usage !
~~~~~ tbc... ~~~~~
Sensors' Behavior in Long-Expos. / Low-Light
Canon was the 1st brand to mount CMOS Image Sensors in its DSLRs in 2002. They're often cited as references for Low-Light, Long-Expos, AstroPhotography. That was true until 2007 when Nikon then Sony came to CMos with Less Noisy sensors than Canon and totally false since Sony's Revolutionary Exmor sensors in 2010. A set of CMos Cameras ( except Aptus75s CCD ) from site P2P.net, Sorted by their Conversion-Noise level, cf PhotonsToPhotos-Charts-Sensor-Characteristics link below
Click Image to Enlarge
Highest Full-Well-Capacity, Q-E, ISO-Invariant values of the Nikon D5, a DSLR Optimized for High Quality images in Very Low-Light + High-Speed conditions ! the Lowest Conversion Noise values to the cameras with Sony Exmor Sensors. Twenty times, 20x lower than EOS-6d's, best Affordable Canon for this Criteria. Pentax K-5 and Nikon D7000 show a near to Null Conversion-Noise value due to their On-Sensor Embedded ADC - Analog to Digital Converter that prevents aSaP the Signal to be polluted by Weak Filtering Power Supply and / or Weak Shielding that will totally ruin their Potential Dynamic Range : Pentax K-5 the TOP Dynamic Range compared to 4 High-End 24x36 DSLRs
Bill Claff, about Photon Transfer Curve in his DPR post # 64718633 : « the single most useful chart showing sensor performance is a pixel level chart » I think you should read the whole thread... and maybe more than once ? DPR PTC Posts :
Bits: useful to know at 1st because all ADU numbers Ranges depend on the bit.depth. If bit-depth =12, then Max ADU(DN) value is 2^12 = 4096 else if bit-depth =14, then Max ADU value is 2^14 = 16384 and Noise values vary Proportionally ! Clicking once on 'bits' you get the 10 bit smartphones, click once more and you get 16 bit CCD cams with high Read Noise Numbers. Older or Cheaper cameras are mostly 12 bit depth, the higher the bit-depth, the stronger you can torture yer Raw files w/o getting Posterization.
Bits: useful to know at 1st because all ADU numbers Ranges depend on the bit-depth. If bit-depth =12, then Max ADU(DN) value is 2^12 = 4096 else if bit-depth =14, then Max ADU value is 2^14 = 16384 and Noise values vary Proportionally ! Clicking once on 'bits' you get the 10 bit smartphones, click once more and you get 16 bit CCD cams with high Read Noise Numbers. Older or Cheaper cameras are mostly 12 bit depth, the higher the bit-depth, the stronger you can torture yer Raw files w/o getting Posterisation.
FWC, Full Well Capacity e- = how much electrons e- stores One Pixel when clipping, the Higher the number, the later Clipped High-Lights will occur. Clicking once on FWC, you get tiny Pixels, CCD cameras, click once more and you get FF cameras with Fewer Big Pixels : their Higher FWC value is fully expected = divide this by the 2 Noise values. Don't be fooled by few High-FWC Big-Pixel, a High Good-Pixel Count is valuable too ! Explanations for DxoMark Per Pixel Screen Mode vs Normalized to 8MP Print Mode,
Read Noise e- : all Noises produced at Pixel level. The lower, the better. Clicking once on RN you get cameras with modern sensors, click once more and you get CCDs, older CMos cameras with Huge Read Noise levels to avoid.
Conversion Noise e- : all the Noises produced by Surrounding Electronics : Analog to Digital Converters, Amplifiers, Power Supplies, etc. The lower the better. Clicking once on CN you get cameras with Modern Electronics, with On Sensor ADC, etc, click once more and you get mainly Canon cameras with their ® Huge Electronic Noise ! 3 Recent models, R5, R6/1Dx3 escape from this totally unbelievable disaster, 12 bit can't fool U !
QE % : Quantum Efficiency = how much incident Photons % are converted to electrons. One Click you get lower QE sensor, CCDs, older CMos cameras with low Fill-Factor, one more click you get the best QEs, iow FSI sensors w/ high Fill-Factor / BSI sensors.
