hello again,
I was testing again today my new canon 50D.
Now I'm showing one simple comparison between Canon 5D and Canon 50D
I'm not an expert doing that, and I know when someone do this, get lots of
complaints, so I'm telling you that I did this just for myself, but as I know
there are lots of people waiting for feedback about the Canon 50D, here you have it.
My goal here is to maximize handheld quality, so ISO 100, canon 70-200/4L IS at f5.6, using IS, and adjusting focal to get the same picture from the same spot (my window)
Canon 5D:
Raw, developed with DPP, Daylight, Standard, Noise Reduction to 0, Sharpness:4
Canon 50D:
Raw, developed with DPP, Daylight, Standard, Noise Reduction to 0, Sharpness:5
Open the TIFF in photoshop, reduce the size to match the canon 5D, and apply a slight Smart Sharpening (30, 0.3, lens blur)
csd2020 wrote:
I am finding more detail in the 50D files and I'm quite pleased with that. It's better than I expected. That was not the case until I started converting raws in ACR. DPP seems to apply some NR no matter the settings at the expense of detail.
Your findings regarding DPP come at no surprise because I fully expected DPP to emulate the in-camera noise reduction algorithm (or something even better) that was used to create the JPEG output files. What I had hoped for was that Canon had "somehow" developed a noise reduction algorithm vastly superior to that of Neat Image without compromising high spatial frequency detail.
Well, the first pre-production 50D sample shots at high ISO had me drooling; they were definitely "at least" 1 stop better than my 40D in-camera JPEGs. Problem was, without one-to-one comparisons to the 40D it was impossible for me to know those 50D pre-production samples were excessively soft. I'm a little down right now ... the drooling has stopped ... but I'm still anxiously awaiting my 50D to show up at the front door.
Back to my basic issue, AFAIC the baggage of extra pixels is counter-productive "if" a compromise of resolution becomes necessary to manage noise.
It's always hard to come up with a meaningful analogy (and I'm lousy at it) but here goes anyway ... Canon claiming the 1.5 stops improvement in high ISO noise and not telling me image detail must be blurred away to achieve said improvement is analogous to a car manufacturer telling me the top-end speed of their latest model is 2.5 times faster than the old model and all the time neglecting to tell me "only if" the new model goes downhill while the old model goes uphill.
I know, I know, I know, my analogy stinks ... I told you I'm not good at that sort of thing ... but it's the best I can come up without thinking about it for two years. I'm now going to retreat from my computer to escape all the "analogy bashers" out there in forum land. :-)
csd2020 wrote:
It may be the increase in resolution or just the micro-adjust capability optimizing my lenses, which once again, makes apples to apples comparisons almost impossible since that feature is not present on the 40D.
The static AF accuracy of my 40D is nearly perfect; therefore lacking MA on it will not hold me back from making a good one-to-one comparison to the 50D. Also, it seems to me you could mitigate AF issues on your end by stopping down around f/8 to f/10 (or resorting to manual focus), just a thought.
Well, I enjoyed this conversation. I wish you the best of luck with the new camera and am positive you will get many great shots with it.
gml1 wrote:
No, I haven't. But empirical evidence is mounting every day.
Here is one example: a guy from DPreview is claiming that the 50D is as good as the 5D - and then he posts these ISO-800 shots (converted from RAW).
Just take a look at the shadow noise (and be honest with yourself ):
I hate to break it to you, that is NOT Empirical Evidence!! I wouldn't use DPR as a standard for anything!!
Go ahead and do a test!! When Joe K gets his 50D, we are getting together and testing it at all ISO's against a D300, Independently processed and we will see. But until someone does that test, the rest is just DPR BS!!
ebe5000 wrote:
hello again,
I was testing again today my new canon 50D.
Now I'm showing one simple comparison between Canon 5D and Canon 50D
I'm not an expert doing that, and I know when someone do this, get lots of
complaints, so I'm telling you that I did this just for myself, but as I know
there are lots of people waiting for feedback about the Canon 50D, here you have it.
My goal here is to maximize handheld quality, so ISO 100, canon 70-200/4L IS at f5.6, using IS, and adjusting focal to get the same picture from the same spot (my window)
Canon 5D:
Raw, developed with DPP, Daylight, Standard, Noise Reduction to 0, Sharpness:4
Canon 50D:
Raw, developed with DPP, Daylight, Standard, Noise Reduction to 0, Sharpness:5
Open the TIFF in photoshop, reduce the size to match the canon 5D, and apply a slight Smart Sharpening (30, 0.3, lens blur)
Thanks for the first brick wall comparison test of 50D.
Can't say I see any differences between the two. If anything the 50D image seems to have a little more details. It's probably safe to say 50D can match 5D IQ pretty closely.
csd2020 wrote:
Many (like you) don't even have the camera yet. I do.
I'll repeat. With the in camera noise reduction features of the 50D, you can easily achieve the 1-1/2 stop improvement in noise over the 40D. You will loose some detail in the process, but you can decide what's acceptable.
Please, stop trying to pass yourself off as an expert based on internet dribble from users with no experience with the camera. I was disappointed at first too but am gaining respect for the 50D with use.
at the same time, we can imagine that the 15MP are outresolving the lens that was finely working on your 30D or 40D. Now imagine you are using some noise reduction on this 15MP picture, in the proces you can't lose much more detail, but you'll clean all the noise, and then, reduced to 8 or 10MP, you have now a 3200 iso picture, without noise, and without detail lost compared to your last camera.
I was thinking that canon should have done a 20MP sensor but for a camera announced for 10MP, because the camera would use the extra resolution to be able to use noise reduction without losing much details and then reducing the picture to 10MP, so the final result for a 10MP picture would be outstanding, but that's just theory.
I'm saying that because 8MP is very fine for me, and i'm not bothered in having a 15MP camera for producing 8MP pictures with huge IQ
IMHO, the 5D pic is sharper even though you have applied more sharpening to the 50D pic (4 vs 5 during conversion, downsized, and used some Smart Sharpen), most easily noticable in the frosted glass window at the top center of the 100% shots. Even though you have used Canon's best zoom for sharpness, the smaller sensor size of the 50D pic is harder on glass, and you can also see a faint purple edge on that same window where the frosted glass hits the metal frame on the bottom. If you can't see it on this example, pull it into photoshop and look at it 200% and check the pixel values on the bottom edge of the window, clearly more red on the 50D sample by around 20 points.
carlk wrote:
Thanks for the first brick wall comparison test of 50D.
Can't say I see any differences between the two. If anything the 50D image seems to have a little more details. It's probably safe to say 50D can match 5D IQ pretty closely.
I don't think you will find big differences on ISO 100 samples from any camera.
IMHO, the 5D pic is sharper even though you have applied more sharpening to the 50D pic (4 vs 5 during conversion, downsized, and used some Smart Sharpen), most easily noticable in the frosted glass window at the top center of the 100% shots. Even though you have used Canon's best zoom for sharpness, the smaller sensor size of the 50D pic is harder on glass, and you can also see a faint purple edge on that same window where the frosted glass hits the metal frame on the bottom. If you can't see it on this example, pull it into photoshop and look at it 200% and check the pixel values on the bottom edge of the window, clearly more red on the 50D sample by around 20 points.
John Ferguson wrote:
....If you can't see it on this example, pull it into photoshop and look at it 200% and check the pixel values on the bottom edge of the window, clearly more red on the 50D sample by around 20 points.
If you need to compare at 200% then it doesn't matter, IMO.
With the in camera noise reduction features of the 50D, you can easily achieve the 1-1/2 stop improvement in noise over the 40D. You will loose some detail in the process, but you can decide what's acceptable.
I have also claimed that DPP seems to apply some NR regardless of settings.
I stand corrected
This may be old news since I really don't spend much time with DPP, but the version that came with the 50D has features I have not seen in previous versions (not that I previously had the most up to date version). The previous versions I've used had a Raw Image adjustment tab and RGB Image adjustment tab in the tools popup. This 50D version also has a LENS/NR/ALO tab and I've been confusing some of the settings under that tab (sorry for my ignorance). It seems that there is a setting to adjust optimal lighting that has labels (LOW, STANDARD and STRONG) which exactly match the in camera noise reduction settings. I thought those settings (in DPP) were related to noise and had not seen much difference in noise levels when changing those settings, DUH. Those settings affect lighting. For noise, there are two sliders (one for luminance and one for chroma). They seem to default to midrange values (scale = 0-20). You can adjust them both to 0 to get rid of NR. OTH, I've found that use of very little luminance 2/20 and more moderate chroma 12/20, very pleasing noise reduction can be achieved with little or no loss of detail. FWIW
I have also claimed that DPP seems to apply some NR regardless of settings.
I stand corrected
This may be old news since I really don't spend much time with DPP, but the version that came with the 50D has features I have not seen in previous versions (not that I previously had the most up to date version). The previous versions I've used had a Raw Image adjustment tab and RGB Image adjustment tab in the tools popup. This 50D version also has a LENS/NR/ALO tab and I've been confusing some of the settings under that tab (sorry for my ignorance). It seems that there is a setting to adjust optimal lighting that has labels (LOW, STANDARD and STRONG) which exactly match the in camera noise reduction settings. I thought those settings (in DPP) were related to noise and had not seen much difference in noise levels when changing those settings, DUH. Those settings affect lighting. For noise, there are two sliders (one for luminance and one for chroma). They seem to default to midrange values (scale = 0-20). You can adjust them both to 0 to get rid of NR. OTH, I've found that use of very little luminance 2/20 and more moderate chroma 12/20, very pleasing noise reduction can be achieved with little or no loss of detail.
FWIW...Show more →
Do you have a 40D? If you do could you please shoot two high ISO (e.g. 1600 and/or 3200) shots using both the 40D and 50D (using the same lens), then develop all shots using exactly same default DPP slider settings and post 100 percent crops of a small area that is in focus and contains high spatial frequency information; thanks very much in advance if you can do this.
Note 1 - the shots should be taken off a tripod, aperture around f/10 or f/11, MLU enabled, and use the "10" second timer if you don't have a remote cable release.
Note 2 - if the AF of either camera is not dead nuts, then focus manually (using an Angle Finder C if you own one). OBTW, if you do have an Angle Finder C don't forget to set your diopter setting prior to use; it's easy to overlook this.
Once again, thanks much if you can test both cameras.
jkurkjia wrote:
Do you have a 40D? If you do could you please shoot two high ISO (e.g. 1600 and/or 3200) shots using both the 40D and 50D (using the same lens), then develop all shots using exactly same default DPP slider settings and post 100 percent crops of a small area that is in focus and contains high spatial frequency information; thanks very much in advance if you can do this.
Note 1 - the shots should be taken off a tripod, aperture around f/10 or f/11, MLU enabled, and use the "10" second timer if you don't have a remote cable release.
Note 2 - if the AF of either camera is not dead nuts, then focus manually (using an Angle Finder C if you own one). OBTW, if you do have an Angle Finder C don't forget to set your diopter setting prior to use; it's easy to overlook this.
Once again, thanks much if you can test both cameras.
Sorry Joe, I no longer have a 40D. I only have a 1D2N right now. I could probably use that for comparison if interested. 40D and 1D2N never seemed much different (ISO wise) to me.
The noise sliders also do not default to midrange values as I previously stated. They default to a value based on the noise level in the image. The higher the ISO (noise level) the higher the slider default. ISO 1600 resulted in a midrange default values. Put another way, the higher the ISO, the more noise reduction automatically applied by DPP. Maybe this is how the 1 - 1.5 stop noise reduction is claimed.
michael49 wrote:
If you need to compare at 200% then it doesn't matter, IMO.
I can easily see the purple and the difference in sharpness at 100% but not everyone has a large monitor nor acute vision ... can you see it? Viewing it at 200% is only a recommendation for those who can't easily see the difference. That purple would be noticeable on a large print.
Wow, I read your take on the 40d/50d and in particular, I'd like to comment on the focus of the 40d. I considered the focust to be a dramatic improvement on the 20d, and the camera taken as a whole to be on the order of the 10d or 1dmkII in terms of what it means to photography in general.
While I don't disagree with your comments on the d300 (I'm seriously considering one), I think these shots from saturday speak for themselves. Given that the light was horrible, and it was raining most of the time, the fact that this gallery represents about half or a third of the total shots taken, I think its a pretty good keeper rate for a prosumer level body.
I wish I still had my 1dmkII or 1d for sports, but the 40d is definitely passable as a sports camera. I'll be skipping the 50d as it really is an incremental upgrade, but If I was coming from the 20d, I would consider it.
csd2020 wrote:
Sorry Joe, I no longer have a 40D. I only have a 1D2N right now. I could probably use that for comparison if interested. 40D and 1D2N never seemed much different (ISO wise) to me.
The noise sliders also do not default to midrange values as I previously stated. They default to a value based on the noise level in the image. The higher the ISO (noise level) the higher the slider default. ISO 1600 resulted in a midrange default values. Put another way, the higher the ISO, the more noise reduction automatically applied by DPP. Maybe this is how the 1 - 1.5 stop noise reduction is claimed.
Steve, thanks for the kind offer but don't take time do it unless it's a specific point of interest to you.
FWIW, and this sort of mirrors your experience, at ISO 800 and 1600 I never found a noticeable difference between my 30D and 1DmkII. If you are interested here are the results (link below) ... a 1DII, 1DIIN, and two different 30D bodies were involved and you need to read the text to see exactly how I normalized the images prior to looking at noise differences.
BTW, when I got my 40D there was a quickie comparison to the 30D and after normalization for file size the noise looked about the same to me; this is not documented at my site because I didn't take care to perform a careful comparison (like I said, it was a "quickie"). Gee, I was just thinking ... was it a quickie that led to Eliot Spitzer's downfall ... no, nope, no way, if memory serves me right it was extended and numerous longies ... so I guess my quickie is okay, right? :-)
Back to the topic, it's a shame you no longer have the 40D but hey, I understand, a buck is a buck, right? :-)
csd2020 wrote:
OTH, I've found that use of very little luminance 2/20 and more moderate chroma 12/20, very pleasing noise reduction can be achieved with little or no loss of detail. FWIW
Thanks for this update... Would you say that when processing the RAW files in DPP this way, it matches ACR/LR regarding the rendering of detail?
Just wanted to say I saw the 50D in the Canon center in MBK mall in Bangkok. It's for 45,700 Baht (convert to your currency using oanda.com). It felt awesome in my hands. Took a couple of shots with it. They looked gorgeous. Although, they were of just the floor. The sales guy mentioned it had arrived only the day before. So, who's gonna be the first to buy it?
Savio_Ax wrote:
Just wanted to say I saw the 50D in the Canon center in MBK mall in Bangkok. It's for 45,700 Baht (convert to your currency using oanda.com). It felt awesome in my hands. Took a couple of shots with it. They looked gorgeous. Although, they were of just the floor. The sales guy mentioned it had arrived only the day before. So, who's gonna be the first to buy it?
Over here (Western Europe - the Netherlands) it is in the stores for almost 2 weeks now. Prices are dropping every week. If you are going to buy one now, you won't be the first for sure
):
