Latest activity

David_B · Sunday at 5:04 AM

AlanF said:
How do you process images? Do you use jpegs straight out of camera or do use RAW? If you use RAW, how do you convert to jpegs or TIFF or whatsoever to view them?

I can see where you're going. What's the point of your question? To try and turn me into a liar because CA/LensCorrection is being done implicitly without me being aware of it? That's a disappointing angle to take.

AlanF · Sunday at 4:45 AM

David_B said:
Let me make your day: I don't use distortion correction when processing images, I can't even remember when I last used CA correction.

How do you process images? Do you use jpegs straight out of camera or do use RAW? If you use RAW, how do you convert to jpegs or TIFF or whatsoever to view them?

AlanF · Sunday at 4:39 AM

David_B said:
IMHO, to do a proper test you'd need to shoot a test pattern chart, the kind that's used by digitalpicture, dpreview, with all the lines at angles to enable measuring lpmm, etc.

As @neuroanatomist has written many times, he has the same charts as the digital-picture and has got them to correct some results. You can do digital-picture type charts for free by downloading them from the bobatkins site. However, to measure lp/mm with precision you need to use IMATEST or similar, which is done by opticallimits, lenstip, ephotozine etc.

David_B · Sunday at 3:35 AM

neuroanatomist said:
It doesn't have to be the same lens, a similar one would be fine.

If someone used different lenses and could prove a difference, would it be accepted? Or would people argue that the results aren't meaningful because different lenses. Maybe it doesn't matter because people will just be argumentative.

neuroanatomist said:
As I've stated, I compared the RF 14-35/4 to the EF 11-24/4, where the latter at 14mm has very little geometric distortion to start. Corner sharpness of the corrected RF lens at 14mm was similar. That supports the idea that digital correction is non-inferior from an IQ standpoint. What I keep asking is for someone to provide some reasonable evidence to back up the claim that digital correction is inferior.

Sorry, I can't help as I don't have the required equipment (a lens that doesn't fill the sensor.)

IMHO, to do a proper test you'd need to shoot a test pattern chart, the kind that's used by digitalpicture, dpreview, with all the lines at angles to enable measuring lpmm, etc.

neuroanatomist said:
Wrong. The distortion correction (stretching) is done during conversion of the RAW image. If you open a RAW image from a lens that doesn't fill the corners and turn off the lens profile, you see the black corners.

Thanks, I didn't know that.

David_B · Sunday at 3:27 AM

neuroanatomist said:
I've suggested before that you refrain from engaging in technical arguments, and you repeatedly fail to heed that advice. Optical correction is bending light, and that is not without consequence. While you are correct that a single photon cannot hit two pixels instead of one, a picture is not generated from one photon. Light from a point in space can certainly be bent in a way that more than one pixel is hit, in fact that's the basis of chromatic aberration, as illustrated for lateral CA here:

While optical correction is bending light, you don't disagree with the statement that optical correction doesn't stretch light. Arguing about bending light is a bit pointless because that's the whole point of a lens - to bend light such that it lands on the sensor.

Some lenses deal with this better than others. Some subjects are impacted by this more than others. Measuring CA is what a lot of lens test websites do when they shoot specific subjects to measure lpmm, etc. Your generalizations here are no better than mine.

neuroanatomist said:
Like geometric distortion, CA is something that is easily and routinely digitally corrected.

What about photon strikes that are in the wrong place? I mean, many ultrawide zoom lenses that fill the corners still have 3-5% barrel distortion. Do you turn off all distortion correction in your images (along with CA correction), so that they remain 'optically pure'? I doubt it.

Let me make your day: I don't use distortion correction when processing images, I can't even remember when I last used CA correction.

mimbu · Sunday at 2:40 AM

DhlcadR6 said:
Not stacked with 40fps isn't a good idea. Unless the 15fps mechanical shutter stays(and no shutter shock), I rather take lower MP stacked CMOS.

The chances of this camera getting a fully stacked sensor are effectively zero. Stacked sensors are hard to make and require fabbing separate layers that are then literally physically stacked on top of each other. Canon couldn't even be bothered to give the R6 III a non-stacked BSI sensor, I'll be surprised if they go to that effort for the lower price & higher volume R7 II. Another ~33mp FSI sensor seems most likely.

rombe3jr · Saturday at 11:43 PM

A bit confused by this. If it is supposed to round out the trinity of STM lenses (16-28 & 28-70) then its overlapping and limited on the long side, at least compared with its competitors in other mounts. For example the Tamron 17-28 f/2.8 ($700 USD on the e-mount), the 28-75 f/2.8 ($700 USD on the e-mount), and 70-180 f/2.8 ($1,000 USD on the e-mount), to round out the trinity with very little overlap. Not sure I like how much overlap a 50-150 would have with the 28-70 and the 30mm short of 180. Sony did release a 50-150 but it is f/2.0.

rombe3jr · Saturday at 11:18 PM

Is it important to have a fast sensor to prevent rolling shutter? I am not a bird / sport shooter but would assume this camera could be useful to that market but with fast moving subjects it seems like a faster sensor would be better.

DhlcadR6 · Saturday at 11:03 PM

I need this lens ASAP

ISv · Saturday at 7:36 PM

Tried today to find again the Orange-cheeked Waxbill. The grass was still wet and there were no birds feeding there. Later small flock (~10 birds) of Common Waxbill came but my target wasn't there: I always have seen it in bigger flocks (like ~40-60 birds). Took bunch of photos of two subadulds Saffron finch in different stages of transition to adult plumage. The second photo is a crop 1580x1579 pixels. Park in Waikiki and the birds are customized to people. I crawl back and they follow me

!!!

Uneternal · Saturday at 7:13 PM

A 15-70mm 2.8-4.0 IS USM would be awesome. I could finally say goodbye to my Sigma 17-70mm 2.8-4.0.

ISv · Saturday at 7:03 PM

AlanF said:
I've wanted to see a Firecrest for a very long time. Just seen my very first one. It was there for just a few seconds and I had to shoot before adjusting settings. These are tiny crops from the centre at iso 51200, denoised as standard in DxO followed by a run through Topaz. Terrible shots but great for me for the record.

View attachment 227430 View attachment 227431

Lucky you! I still have no photo of the Firecrest. Goldcrest - yes (few miserable photos...).

PinholeR5 · Saturday at 4:48 PM

P-visie said:
The length in the patent application is from the front of the lens to the image plane, so it includes the 20mm flange distance. So you need to substract 20mm from the lens length in the patent application. The lens size would be almost the same as the RF70-200mm f2.8.

See: https://www.canonrumors.com/canons-new-patent-ultra-compact-body-cap-lenses-for-the-rf-mount/

Ah, cool, thanks! So I was missing something

Still, it would be marginaly larger than the RF 70-200mm F2.8 and meaningfully larger than the F4.

TexPhoto · Saturday at 4:03 PM

Before an R3M2 I'd like to see an R7M2 that was actually an evolution of the 7D series. Metal body, etc... I don't think we will see either an R3 or metal R7.
But if we did get an R3M2 prediction:

1. FF 30MP Global Shutter.
2. 50 FPS with 2 seconds of precapture.
3. Noise performance of a EOS 10D. (Kidding mostly)

neuroanatomist · Saturday at 3:57 PM

David_B said:
Lets get it right: optical correction is not stretching. The goal with optical correction is to try and ensure that as many individual photons are distinguishable as possible when hitting the thing capturing the image. You can't stretch a photo to make it hit two pixels instead of one, but what you can do is try and ensure that two photons on very similar trajectories hit individual pixels on the sensor.

With firmware/software stretching photon strikes that didn't happen have to be invented.

I've suggested before that you refrain from engaging in technical arguments, and you repeatedly fail to heed that advice. Optical correction is bending light, and that is not without consequence. While you are correct that a single photon cannot hit two pixels instead of one, a picture is not generated from one photon. Light from a point in space can certainly be bent in a way that more than one pixel is hit, in fact that's the basis of chromatic aberration, as illustrated for lateral CA here:

Like geometric distortion, CA is something that is easily and routinely digitally corrected.

What about photon strikes that are in the wrong place? I mean, many ultrawide zoom lenses that fill the corners still have 3-5% barrel distortion. Do you turn off all distortion correction in your images (along with CA correction), so that they remain 'optically pure'? I doubt it.

neuroanatomist · Saturday at 3:32 PM

David_B said:
It can't meaningfully be done with the equipment we have available because you can't use the same lens to do both at the same time, meaning that there will always be differences in the image created.

The best you could do would be to take a lens such as this and create a system where you can alter the position of the sensor relative to the back of the lens so that when it is at 50mm, the position of the sensor is further back allowing the spread of light to cover more area. that assumes that the photos coming out the back of the lens are coming out at an angle.

It doesn't have to be the same lens, a similar one would be fine. As I've stated, I compared the RF 14-35/4 to the EF 11-24/4, where the latter at 14mm has very little geometric distortion to start. Corner sharpness of the corrected RF lens at 14mm was similar. That supports the idea that digital correction is non-inferior from an IQ standpoint. What I keep asking is for someone to provide some reasonable evidence to back up the claim that digital correction is inferior.

David_B said:
It would also require custom firmware to record what comes out of the sensor because I think Canon do the stretching very early so you can never see an unstretched image.

Wrong. The distortion correction (stretching) is done during conversion of the RAW image. If you open a RAW image from a lens that doesn't fill the corners and turn off the lens profile, you see the black corners.

This is the corner of a shot with the RF 24-105/2.8 Z taken just after I got the lens, before DxO had a correction profile.

Incidentally, there's may be an easier way to achieve a practically meaningful result – the RF 24-105/2.8 Z doesn't fill the corners at 24mm but does by 28mm (maybe wider, I have not checked). So a comparison of filling the corners with digital extrapolation vs. with light is possible using the same lens (albeit at a different zoom setting, but a close one). I may try that at some point.

Tomba · Saturday at 3:15 PM

Sporgon said:
I checked it out; some terribly nervous bokeh at f/1.2, the boy on the green slide for instance, looks like the RF 50/1.8 !
The planar optical formula does tend to do this; adding an aspherical element makes it even worse. Even the EF 50/1.2 could get nasty. Kind of makes me confirm that if you want shallow dof and tasty bokeh with a ‘normal’ lens use an 85 or 100.

As I said, it's not at RF-L level. Only EF 50mm 1.2 L level. If you know the EF 50 L and like the look you will like the the RF 45. If not then RF 50 1.4 L VCM or RF 50 1.2 L will satisfy your needs.

Click · Saturday at 2:56 PM

I really like the abstract aspect of the last one. Well done, Nemorino.

Sharlin · Saturday at 2:55 PM

Maximilian said:
With the 28-70 STM ending at 70 mm, this 70 mm should be the starting FL and it should go further than 150 mm.
With 200 or else preferred.

You mean Canon doesn't already have 70–200mm RF lenses?

In any case, if this ever becomes reality and is reasonably priced, it would be a perfect partner to a hypothetical, sorely lacking 17ish-50ish mm f/2.8 RF-S. Not every lens needs to be a part of a FF "trinity".

Del Paso · Saturday at 2:53 PM

Snapster said:
Maybe a travel lens to combo with the 14-35 L. Maybe.

Probably still would rather take 70-200 f4 and a fast 50mm prime. Actually this.

That's "exactly" what I do, for day-trips, RF 15-35 f/2,8 + RF 70-200 f/4, each on its own body. I dislike changing lenses (missed occasions and laziness).
But for longer trips, I add the RF 100-500 + EF 24mm TSE + 60mm macro.

P-visie · Saturday at 2:52 PM

PinholeR5 said:
At first look, my feeling was the same. This lens would complement Sigma's 10-18mm and 18-50mm very nicely. But it seems to be too large. I would actually expect (or, perhaps hope) that Sigma would come out with something more like a 50-115mm F2.8 that is more compact.

When I look at the table in the article it lests the lens length as:

176.15 (W) and 236.15(T)

I'm assuming these are in mm. When I look at the spects for the RF 70-200 F2.8 (non-Z) & F4 it lists the lens length as:

F4 : 119 (W) and 181(T) (https://www.the-digital-picture.com/Reviews/Lens-Specifications.aspx?Lens=1529)
F2.8 : 146 (W) and 212(T) (https://www.the-digital-picture.com/Reviews/Lens-Specifications.aspx?Lens=1417)

Which suggests that the 50-150mm would be longer than both of the current RF L non-z 70-200mm. Am I missing something? Is it just the diameter and weight improvements? Lower price? More versatile focal length range? It really doesn't seem to complete the "non-L f/2.8 trinity".

The length in the patent application is from the front of the lens to the image plane, so it includes the 20mm flange distance. So you need to substract 20mm from the lens length in the patent application. The lens size would be almost the same as the RF70-200mm f2.8.

See: https://www.canonrumors.com/canons-new-patent-ultra-compact-body-cap-lenses-for-the-rf-mount/