Er Small Raw i Canon = binning = øget følsomhed

[michaelrDeltager Super Giant]

Hej

Er der mon nogen der ved om det nye Canon format på 50D med Small RAW eventuelt svarer til binning og øget følsomhed?

Teksten nedenfor er klippet fra http://www.northlight-images.co.uk/cameras/Canon_50D.html

“Two Small RAW Formats
In addition to the RAW and JPEG image capture modes that photographers are accustomed to, the EOS 50D Digital SLR camera now offers more manageable resolution settings and file sizes with two sRAW recording formats, sRAW1 and sRAW2. At the sRAW1 setting, resolution is 7.1 megapixels with a file size that is approximately 25 percent smaller than a standard 15.1 megapixel RAW image. With the sRAW2 setting, resolution is 3.8 megapixels at less than half the file size of a standard RAW image, retaining all of the flexibility and creative possibilities associated with full-size, conventional RAW images. The EOS 50D is the first Canon Digital SLR that allows the use of RAW and sRAW settings in Basic Zone as well as Creative Zone shooting modes, even further improving the camera’s flexibility.”

[michaelrDeltager Super Giant]

Fandt dette som vel giver et ½ svar, at binning er på vej for CMOS, men at Canon generelt er et stykke bagefter Nikon mht støj.

http://74.125.77.132/search?q=cache:WZ3yWK_hem0J:www.nikonians.org/forums/dcboard.php%3Faz%3Dshow_topic%26forum%3D226%26topic_id%3D2981%26mesg_id%3D2981%26page%3D+binning+50D&hl=da&ct=clnk&cd=9&gl=dk

“Pixel binning is a data accumulation technique that was originally designed specifically for use with CCD sensors. Digital camera CCDs can be designed to clock multiple pixel charges in both the horizontal and vertical direction into a single larger charge sometimes referred to as a superpixel. The superpixel represents the area of all the individual pixels contributing to the charge. The creation of a superpixel is called binning because multiple pixels are stored in a single electronic container or bin essentially. According to various technical papers, binning of 1×1 means that the individual pixel is used as is. A binning of 2×2 means that an area of 4 adjacent pixels have been combined into one larger pixel, and so on.

The CMOS sensors used variously in the 5D, 50D and the D700 don’t suffer from the same electrical noise issues typically found in charge-coupled (CCD) sensors.

Binning techniques in CMOS sensors include reducing the transistor-to-pixel ratio, standard signal averaging, and other methods.

All the techniques are perfectly good approaches whether implemented by Kodak, Sony, Canon or Nikon. Unfortunately, the physics and electronic engineering involved eventually always bump their heads on the physical size of the APS-C sensor as well as the practical limits of current processing power in camera CPUs. While the CPUs are perfectly capable of handling binning calculations, designers and engineers are forbidden from occupying all of the CPU’s power with one task (such as overly complex pixel binning and data arbitration) because of all the concurrent tasks a camera has to perform every time someone presses the shutter in continuous focus, multiple frame, 3D tracking mode.

Every technological base in electronics has it limits and I think the 50D stretches things to just before the breaking point. Given the fact that engineers at Canon and Nikon are equally brilliant, and given the natural constraints of APS-C sensors compared to full frame/FX sensors, there is no practical/affordable way for the APS-C CMOS sensor to come anywhere close to the low light/low noise/high ISO performace of the FX Nikon sensor – at least not without driving the price of the process through the roof and beyond even the lofty heights already achieved by goofy-expensive cameras such as the 1Ds MKIII and the D3.

The 50D’s APS-C CMOS barely scratches at the mid-point of the positively ancient (though still superb) full frame 5D CMOS high ISO noise performance (keeping in mind of course that people raved about the 5D’s superb low noise performance at ISO800 – nowhere near the remarkable low noise ISO3200 performance of the D3 and D700). Nikon’s accomplishment in this area is truly spectacular. That pros are regularly shooting, publishing and selling money shots with the D3 at ISO6400 is just plain wierd. So after all this time, Canon has rev’d the prosumer/semi-pro series twice and still hasn’t bested the wonderful, pro, 5D in image quality or noise performance. Nikon, on the other hand, has put more resources into the much more cost effective effort of scaling the full frame CMOS into a something with four or five times better low noise/high ISO performance than the 5D along with almost equal image quality across the board (and remarkably better image quality in marginal and low light).

S’my view on the subject. While Nikon’s genuises may be working on ever more exotic pixel binning for an APS-C successor to the D300, they’re also working on even better performance with FX sensors. I also firmly believe that Nikon is heading for full frame exclusively in its SLR line over the next five to six years, gradually sunsetting all of the APS-C sensors. I think we’re shortly going to see a whole (rather nice) range of superior point & shoot cameras (from Nikon, Canon and Sony) containing APS-C sensors.

I’m delighted with my D3 (since last winter) and D700 both of which I’m using now to shoot what I want, whenever I want, and if necessary in lighting conditions which, as recently as a year ago, would have forced us all to call it a day and head for home.”

[michaelrDeltager Super Giant]

Jeg fandt disse 3 optagelser i hhv RAW, SRAW1 og SRAW2, alle ISO 3200. Så vidt jeg visuelt kan se er der noget mindre støj på SRAW1 og endnu mindre på SRAW2, men er der nogen der kan måle på dem? Fundet på http://www.pbase.com/csd_5092/stuff__misc&page=2

http://rask.siriusastro.com/images/SRAW/104058811_q2gXbC4X_raw1.jpg

http://rask.siriusastro.com/images/SRAW/104058815_Kkd3gfRj_sraw1.jpg

http://rask.siriusastro.com/images/SRAW/104058816_CFTnqBCf_sraw2.jpg

[morten.brixDeltager Super Giant]

Hej Michael

Jeg tror bestemt ikke at SRAW giver en øget følsomhed.

1. Så tror jeg at canon ville fortælle omverden om det.

2. Digitale kamera er afstemt iforhold til ISO værdier.

3. Når man laver binner bliver billederne sort/hvide.

Det kunne være at Jacobb el. Jesper Grønne kunne kaste lidt lys over denne sag.

[ac Planet]

Hej Michael

Prøv at kikke forbi på http://www.astrosurf.com/buil/

Han har netop smidt en sammenligning mellem Canon 40D og 50D på sin side, og han har også en sammenligning mellem Canon og Nikon.

Det har hidtil været sådan, at Canon har valgt en tilgang, hvor støj fjernes ved at mindske generering af støj i CMOS’en og den øvrige elektronik, mens NIKON har fokuseret på at mindske støjen ved brug af et median filter. Nikons filter kaldes også “stjernespiserfilter”, da filteret ud over at fjerne støjen (såvel termisk som bias/udlæsningsstøj) også fjerner svage stjerner som kan forveksles med støj.

Til dagslys skulle Nikon være bedst, mens Canon foretrækkes af amatørastronomer da de ikke spiser stjernerne. er iøvrigt enig med brix mht binning.

Buil har mange gode artikler på sin side, så god læsning.

mvh AC

AC2008-12-06 17:44:14

[Anonym Nova]

Er der nogle som har sammenlignet “hardware” vs “software” binning med deres CCD? Umiddelbart skulle man jo tro at udlæsnings-støjen er meget lav iforhold til andre støjkilder (pånær evt ved narrow-band billeder), så man derved ikke vinder det store ved binning. Men er der nogle som har testet dette i praksis?

Man kan måle støjen i et billede ved at hente det ind i IRIS og skrive “stat” i kommandoprompten.

Søren

[michaelrDeltager Super Giant]

Tak for linket til Buil, det er spændende læsning. Uheldigt at 50D har mere tilfældig støj. Desværre skriver han ikke noget om binning i DSLR. Som jeg læser teksterne tror jeg i tager fejl, venner. Men jeg vil forsøge at få Delfer til at tage nogle darks med hans Nikon D700, det kan vist også binne, eller hvad det nu er for noget i DSLR sammenhæng.

Har prøvet at køre de 3 billeder igennem IRIS og median tallene for billederne i den rækkefølge de står ovenfor for RGB er som følger, Søren hvad udtrykker det? Mindre støj:

R 86 75 73

G 94 84 83

B 87 82 71

[motomandk Main Sequence]

Er der nogen der har et link til en sammenligning mellem 30D og 40D?
Det kunne være sjov at se om der er en sammenhæng mellem de stadig mindre pixels -> mindre følsomhed -> højere gain (for at opnå samme ISO) -> mere støj.

/Henrik (H)

[joergen.kDeltager Nova]

her er et billed iso 1600 10 min ved stuetempertur 21grader

uden støjreduktion.1024×683

hvis der en med et eos 30 der kan tage et billed med samme data kan vi sammenligne

/jørgen k

[joergen.kDeltager Nova]

her samme data med støjreduktion

[michaelrDeltager Super Giant]

var det en ide at vi sammenlignede vores DSLR kameraer mht. støj. Vi kunne hver lave 2 darks ved stuetemp og 1600 ISO (eller bør det være 800?) hhv 30 sekunder som kan klares uden bulb og 300 sekunder som der sikkert er mange der bruger til DSO. I RAW og gerne RAW1 og RAW2, hvis der er folk der har det (Canon 50D, måske også Canon Full Frame og visse af Nikon Full Frame)? Måske ville MS modtage billederne, analysere for støj og hoste (jeg kan også, men ikke hoste) ?

[michaelrDeltager Super Giant]

Hov, glemte at nævne at det nok ikke skal være støjbehandlet og heller ikke reduceret.

Og til Henrik, ja der er en klar sammenhæng, hvilket Buils sammenligning mellem 40D og 50D også viser. Men den er ikke proportional.

[joergen.kDeltager Nova]

ok bare sig til den er jeg med på

/jørgen k.

[mstauningDeltager Black Hole]

Det er jeg også i sender bare.

[michaelrDeltager Super Giant]

Hej Michael

Så er der sendt 2 darks i RAW mærket Canon400D800ISO30sek hhv …300sek

Ud over et par målinger eller kurver eller hva du finder på, laver du også en ens stretch, så man kan se eventuelle forskelle visuelt? Det er da spændende!

[Forfatter]

Indlæg