Canon registers a 32.5mp APS-C DSLR in Taiwan

Mar 26, 2014
1,443
536
I have made it absolutely clear that I am a member of group that wants high megapixel sensors because they aid our type of photography. If you consider that as a "bias", then you are biased also.

I'm biased, you're biased, and that doesn't say anything concrete about 'one in how many photographers benefits from cameras that have sensors with more than, say, 50MP'.
 
Upvote 0
Mar 26, 2014
1,443
536
... or the megapixel race towards ~ 20Mpix modes which have a RGGB quadruplet as a base for a non-bayer full color pixel as alternative mode to save moderate resolution high color depth images while retaining DPAF at a very high granularity. + enable smaller AF "points".

This idea has been raised before. Good point.
 
Upvote 0
Mar 26, 2014
1,443
536
Okay, I will be more direct. Why are you doubling down?

Okay, I'll repeat myself. I didn't.

Maybe 99% of people don't need 32 mp sensors. So what?

So I wonder why the subject of >30MP sensor, e.g. 83MP FF sensors, gets raised so often.

I get that years ago DSLRs had such low resolution, most people needed more, and raises in resolution were often discussed. I don't get why its still so often discussed. This thread doesn't count, people just enjoy arguing with me.

Digital cameras have become a niche market and Canon's path to success/survival is serving that niche market.

My guess said path is served by making items that sell well, but what do I know?

You say your point is that the number of posts by people who want more megapixels is disproportionate to the number of people in the general population that need or want more megapixels. So what? This is a thread about a sensor with more megapixels, on a geek forum devoted to new camera technology and read by the less than 1% of the population who are the actual targets of these products. Why shouldn't it generate interest among those who want more megapixels?

I suspect those people are 1% of the people who participate in this geek forum.

BTW, given your insistence on holding on to an idea that others have decisively shown to be invalid, I have to ask, did you used to post under the name Dilbert?

I never held that idea, it's words people put in my mouth because they enjoy arguments.
 
Upvote 0
Mar 26, 2014
1,443
536
We're an art gallery and print shop. If you're printing concert quality posters, all files are preferred to be at 300DPI. You can easily get by on small res files, but it's far from ideal. Especially for large prints. I've worked with a guy like you before.

No, you didn't. That guy only saves files in 72DPI, while I would save to whatever resolution my boss would pay me to save it, regardless of my opinion on the matter.
 
Upvote 0
Oct 22, 2014
168
126
No, you didn't. That guy only saves files in 72DPI, while I would save to whatever resolution my boss would pay me to save it, regardless of my opinion on the matter.
That doesn't make any sense. Why save a low quality file unless you're sending a specific copy of said file off to print? Like, if you're just printing a window poster, 72DPI is passable, but what does money have to do with saving the original file at it's max resolution? We must work in two entirely different industries. Nothing is ever below 300DPI unless a customer brings in a cell phone image, which are inherently 72DPI.
 
Upvote 0

dtaylor

Canon 5Ds
Jul 26, 2011
1,805
1,433
... or the megapixel race towards ~ 20Mpix modes which have a RGGB quadruplet as a base for a non-bayer full color pixel as alternative mode to save moderate resolution high color depth images...

Bayer gets a bad rap. There really isn't much to gain in terms of color depth or resolution by eliminating Bayer. Foveon images show an improvement but a small one that's not in line with the exaggerated advertisements. In terms of an 80mp sensor with a 20mp 'color depth' mode I doubt there would be any gain after scaling the normal 80mp version down to 20mp and comparing with the color depth 20mp one.

Honestly, Fuji solving the moire issue by alternating the CFA arrangement seems a better improvement than triple layer sensors (Foveon) or a theoretical 4:1 sensor. Of course that leads to the issue that so much optimization has been put into Bayer processing that Fuji images can suffer during RAW conversion.

Now I wouldn't refuse a triple layer sensor if Canon offered a high resolution one that didn't suffer at high ISO. There is some improvement. But not one I would trade resolution or high ISO for.
 
Upvote 0

jolyonralph

Game Boy Camera
CR Pro
Aug 25, 2015
1,423
944
London, UK
www.everyothershot.com
I'm still amused about how many people argue about printing posters when large resolution sensors are discussed.

It's not about printing posters. It's about increasing range through cropping, and allowing downsampling to reduce the problems with bayer filters.

Thank you!
 
  • Like
Reactions: 1 users
Upvote 0
Mar 20, 2015
428
372
It's not about printing posters. It's about increasing range through cropping, and allowing downsampling to reduce the problems with bayer filters.

You're generalising your experience.

Just yesterday I was working on a 60x40 print based on a 1Ds3 shot ( so 20MP ). There was just enough resolution to achieve it, because of the composition and subject. But don't look too closely at the dog's hair or his eyes.

So I could say that it's all about printing posters. Neither of us would be correct. What is correct to say is that most photographers will always find a use for more MP.



Now, can we get back to this camera? Anyone fancy a guess at release date?
 
Upvote 0
Bayer gets a bad rap. There really isn't much to gain in terms of color depth or resolution by eliminating Bayer. Foveon images show an improvement but a small one that's not in line with the exaggerated advertisements. In terms of an 80mp sensor with a 20mp 'color depth' mode I doubt there would be any gain after scaling the normal 80mp version down to 20mp and comparing with the color depth 20mp one.

Honestly, Fuji solving the moire issue by alternating the CFA arrangement seems a better improvement than triple layer sensors (Foveon) or a theoretical 4:1 sensor. Of course that leads to the issue that so much optimization has been put into Bayer processing that Fuji images can suffer during RAW conversion.

Now I wouldn't refuse a triple layer sensor if Canon offered a high resolution one that didn't suffer at high ISO. There is some improvement. But not one I would trade resolution or high ISO for.

Downsampling makes the same quality, for shure. But it can decrease file size maybe by a factor of two. And some color moiree of strong quasi monochromatic light sources would be avoided.

Foveon has a strong problem with ISO of the deeper layers so they are usable and very good at base ISO but fall apart at ISO 400 or ISO 800.

The only way out would be some structure which separates the different wavelength ranges like a grating or prism to use it all - but I do not see a simple solution to manage that with light which comes from different angles.
 
Upvote 0
I speak English as a second language, so the finer points of English spelling are lost on me. Which is why I used the ‰ character, rather than spell it out.

Around here it's taught in elementary school along side percents, after fractions, probably to demonstrate those are just names for "parts per hundred", "parts per thousand", and "parts per ten thousand".

Yes, and after a second I realized the meaning of the symbols with the percent symbol, and I'm glad you added yet another nerdy nugget to my brain. Just do be aware that to the US at least, those terms might as well not exist.
 
Upvote 0

AlanF

Desperately seeking birds
CR Pro
Aug 16, 2012
12,344
22,520
Yes, and after a second I realized the meaning of the symbols with the percent symbol, and I'm glad you added yet another nerdy nugget to my brain. Just do be aware that to the US at least, those terms might as well not exist.
They are not in general use in the UK either. And a perusal of the course requirements for mathematics taken by 16 year olds (GCSE) fails to find the symbols as part of the syllabus.
 
Upvote 0
if they ever do merge the 7D and the 80D they should name it 9D. However, I am not sure how those two very different body options can be merged in a way both type of uses will be happy with them.
Great points!
I've always thought that the 80D successor will need to be renamed to 8D or 9D, as the xxD series is running out of names.
And yes, with a 32.5mp sensor, they can't make both the 7D and 80D camps happy.

Btw, 32.5mp is just a 16% increase in linear pixel density vs a 24mp sensor.
If Canon keeps pixel performance the same as on the 80D, overall image quality will be better.
I had the 40D before and I used to think that we don't need more megapixels.
Now I have the 80D and the images are so much better.

And a final point: the move to a higher resolution signifies the arrival of the next iteration of Canon sensors.
So, again, image quality will (very likely) be better.
 
Last edited:
  • Like
Reactions: 1 user
Upvote 0
===

In terms of our INTERNAL printing of posters, we usually PRINT at 2400 dpi (individual dots per inch) which is about 600 ppi (i.e. CMYK pixel groupings per inch). Since we usually shoot using a 1dx Mk2 for up to 8x10 inch or a 100 megapixel Hasselblad that works out to about 150 ppi if we scale it to 19x13 inches.

Since we ALSO use a FRACTAL SCALER-based rasterizer on our print jobs, the combination of fractal scaling and error diffusion makes our photos look great even when printed out at 48 inches by 36 inches and larger.

THE KEY ISSUE when printing is NOT the actual dpi of the source camera image but rather the QUALITY of your print rasterizer which SHOULD USE Fractal Scaling and/or Stochastic Error Diffusion.

MOST newer Epson or Canon pro-level printers will have 2400 dpi and ERROR DIFFUSION settings which you should turn ON for best photo quality output!

If you have the TIME, using an external software-based image scaler/rasterizer is best, but modern printers now give you 90% of that sort of quality within the printer of itself using the built-in error diffusion algorithms.

One thing I have noticed, is you MUST USE a colour-matching system (i.e. Datacolor SpyderX Elite is one of the best monitor calibrators!) so that your printout will MATCH what is displayed on your monitor within your image processing and publishing program.

Every morning I colour match my monitors and printers for BOTH video and print uses and save a preset so I can switch between the YCbCr video-centric display modes and print-centric CMYK modes. Since monitors have COLOUR DRIFT over a time period as short as even a few hours, I usually do it in the morning.

Not everyone will need that sort of colour accuracy, but by buying even the cheaper Spyder colour calibrators, your printouts will match your monitors pretty closely!

---

And while the point of high resolution is for allowing more cropping options and better print resolution, it's NOT always necessary since MANY publications are WEB-specific which means for 8.5 inches by 11 inches PDF files, you can get away with a full page portrait resolution as little as 2000 by 1500 pixel or about 180 dpi! That means your camera really needs to be at a minimum of maybe 6 megapixels to give you enough room to do a proper downsample and crop for web-based publishing!

In my opinion, two of the BEST cheaper MILC cameras for doing that sort of web-publishing are the Canon M-50 and the Sony A6500. The quality and resolution are MORE THAN ENOUGH for web and the price is right! If you are JUST doing web photos and maybe smaller 1920 by 1080 embedded videos then the $799 Canon M50 is a good start.

If you're doing BOTH print AND vlogging then the $1600 Sony A6500 is much better for that!

.

Got your G7 master cert, homie?
 
Upvote 0

Sharlin

CR Pro
Dec 26, 2015
1,415
1,433
Turku, Finland
So I updated my throughput spreadsheet and found that the jump to 32 Mpix, although large, is not entirely unreasonable. There's even room to slightly increase continuous shooting rate: at 7.5fps the throughput is equal to the max throughput of the M50 (24MP x 10fps), and at 8fps only slightly more. I doubt we'll be seeing much higher number than that, however. Also included a hypothetical 1DX3 at 24MP x 16fps.

184452
 
Upvote 0

Sharlin

CR Pro
Dec 26, 2015
1,415
1,433
Turku, Finland
They are not in general use in the UK either. And a perusal of the course requirements for mathematics taken by 16 year olds (GCSE) fails to find the symbols as part of the syllabus.

They're almost nonexistent in the Anglophone world, but permille/promille at least is somewhat common in many other languages. However I'm not sure I've seen permyriad in the wild before this!
 
Upvote 0
So I updated my throughput spreadsheet and found that the jump to 32 Mpix, although large, is not entirely unreasonable. There's even room to slightly increase continuous shooting rate: at 7.5fps the throughput is equal to the max throughput of the M50 (24MP x 10fps), and at 8fps only slightly more. I doubt we'll be seeing much higher number than that, however. Also included a hypothetical 1DX3 at 24MP x 16fps.

depends. on an M5 that would be a drop in FPS, and then on top of it, the added overhead of DPAF which will cause continuous to struggle even more.

ALso, the M50 only does 12 bit continuous at 10 fps.
 
Upvote 0