9VIII said:
privatebydesign said:
Mikehit said:
neuroanatomist said:
But doesn't it really come down to budget? If your wedding has a budget for a photographer who has a 1D or 5D then that person has the extra card for just in case and that's great but that equipment is bought from their fees. I bet there are plenty of budget wedding photographers who don't even have full frame cameras never mind dual cards and assistants. Not every wedding is an extravaganza but the participants still want pictures of it.SecureGSM said:
scyrene said:
magarity said:
9VIII said:
ahsanford said:The bold bit above is the entire story here. Do we have data to back that up?
I am not doubting you -- it seems a formidable camera -- but do we know it is really kicking Canon's tail or if it is simply bringing the long-neglected Nikon 'pro APS-C' birder/wildlifers back into the fold? Is it stealing market share or is it just a case of shifting faithful Nikonians into a new price point?
We're starved for market data here, so it's hard to peg if:
[list type=decimal]
[*]This rumor is BS
[*]This rumor is true, and Canon is actually accelerating its plans with the 7D3 due to market forces (i.e. the D500)
[*]This rumor is true, and Canon is simply deploying the 7D3 seemingly sooner to us because of the 5 year lifecycle for the 7D1 was an exception*, outlier, etc. and the 7D2 lifecycle is actually more in the line with the other major rigs on a traditional-ish 4 year timetable.
*this pertains to the unique market situation around the whole earthquake / Nikon abandoning the segment without a D300S follow up / the decision to do a firmware 'lifecycle extension' to the 7D1, etc.
[*]This rumor is true, and Canon has found the means/resources to accelerate its development pipeline as it sees the ILC market as having bottomed-out and it believes that now is the time to go big and saturate the market with new bodies. (I'm not buying this at all.)
[*]This rumor is true, but some important new product slated for 2018 had a major delay or problem, and the 7D3 was brought forward to prevent a 'hole' in the pipeline from leaving Canon looking bad next year. (This is wild speculation, feel free to run with this nutty idea
[/list]
Almost all my money is on #1 above, perhaps #2 -- but again, the public never sees market data in any appreciable depth that would be able to verify this.
Curious to see where everyone's heads are on this. What's your guess?
- A
Talys said:9VIII said:
Assuming you mean striped recording, it's factually untrue that striped volumes reduce write performance. In fact, they dramatically increase write performance, and almost every serious server uses striped data sets to improve performance. The easiest way to make either mechanical or solid state drives faster is to buy a bunch of them and split up the write tasks between them.
There are massively expensive RAID setups that make a 1000mm primes look like rounding errors, and buildings full of them that cost tens or hundreds of millions of dollars to build, and a lot of them use mechanical storage for a lot of storage tasks -- for example, Microsoft's Azure or Amazon's AWS datacenters. Some of us also choose mechanical storage on the cloud for non-performance reasons, like better geo-replication options.
In addition, many people who run on-premise, production servers choose mechanical over solid state drives. They're superior in two ways: first, if they die, they tend to die slowly, so you can typically recover data easily. On the other hand, SSDs die in an epic and sudden way that's either difficult, expensive, or impossible to recover data from. Second, unless you get extremely expensive SSDs, very expensive RAID cards, or both, you can't REST them while they are in a striped set -- which makes all those cheap retail drives just short of useless.
And finally, there's limited benefits on a lot of types of servers. For example, for email or database, you're going to cache nearly everything you use most of the time anyways, so those blazingly fast random read times will still be there, but practically, they aren't noticeable much to users because usually, the data they want to access is cached in much faster RAM.
A strategy is often to combine both, and put most of your storage on mechanical storage arrays, but use solid state as a temporary/scratch drive, or for data that you can predict will require extremely random reads that don't cache well.
By the way, mechanical drives have improved dramatically since 2010. You can get excellent 2.5" models now, which are king in the server world. The reason that the price per TB hasn't fallen on mechanical drives has less to do about technology than it does about market forces. There just isn't enough demand for those 12TB systems to bring the prices down. Those giant drives also introduce other problems, like length of time to back up, and how hard you cry if you lose everything.
There have also been massive improvements in SSD technology since 2010, a lot of it highly benefiting retail PCs in the tablet and laptop segments. m.2 SSDs are a big step up in both performance and form factor for drives that don't need to be changed often, the controllers have gotten a lot better, and MLC durability has improved a lot. Basically, mid-range drives today are great, whereas they were slightly lacking a decade ago.
Not that this has much of anything to do with 7D3
9VIII said:
I project a lot of still images and the difference between my 6D and the 5DS are very easy to see, better tones, better clarity and more fine detail.Sporgon said:jolyonralph said:Khalai said:If you print small or do not heavily crop, then why would you need high resolution in the first place, right?
Downscaling.
Remember that a 24mpx sensor doesn't give you 24 colour megapixels. there are 12 megapixels of green, and 6 megapixels each of red and blue.
These are then interpolated to give you a pseudo-24 megapixel full colour image, bearing in mind every single one of those pixels is the product of assumption, taking the surrounding pixels to assume what the missing colour components would be.
So, even if you don't crop, or you don't print massive posters, the 5DSR gives you the opportunity of massive oversampling which allows you to scale down an image and get a much more accurate representation of both the luminosity and color of each actual pixel in your final image.
I just don't see this in practice. I wish I did, it would mean I could justify buying a 5Ds to produce better quality images. Theoretically of course you are right, the bayer array means that you need many more pixels on target to define colours, especially at the edges of things, but I just don't see it in practice.
The problem is all to do with scale; because of the resolution of even a 12 mp FF camera there is enough RGB coverage to define the detail we can see in say an A3 size print. The increase in colour definition that you are gaining from say a 5Ds is just lost in reducing the detail size down to an A3 print. That detail has just gone. Simple. If it hasn't then it was large enough for the lower mp sensor to define anyway.
I have found that the greater magnification of a larger format gives better tone and detail, so for instance again if we refer to a fairly common "large" A2 print (23.5 x 16.5 inches) a vertical three frame stitch from an old 12.7 mp 5D at about 24 mp is better printed definition than a single frame from a 24 mp M3. However as you reduce the print size, once again there is no difference.
Of course cropping is another thing, you are definitely limited in producing large prints from a heavily cropped 12 mp sensor
ahsanford said:
Hard drives have a single head per platter which performs both reading and writing, not one for each function.9VIII said:
The term you're looking for is "sectors" which is the smallest atomic change. A sector may contain multiple blocks or vice versa depending on the formatting. However, sectors are fairly small and most files spread across many of them. When changing a file only the sector(s) being changed need to be rewritten. Most files are not frequently changed in the big scheme of things. Even editing an image or video is a series of single changes but it's not like anyone re-edits the same video over and over until those sectors give out (unless they have some compulsion).
jeffa4444 said:I project a lot of still images and the difference between my 6D and the 5DS are very easy to see, better tones, better clarity and more fine detail.Sporgon said:jolyonralph said:Khalai said:If you print small or do not heavily crop, then why would you need high resolution in the first place, right?
Downscaling.
Remember that a 24mpx sensor doesn't give you 24 colour megapixels. there are 12 megapixels of green, and 6 megapixels each of red and blue.
These are then interpolated to give you a pseudo-24 megapixel full colour image, bearing in mind every single one of those pixels is the product of assumption, taking the surrounding pixels to assume what the missing colour components would be.
So, even if you don't crop, or you don't print massive posters, the 5DSR gives you the opportunity of massive oversampling which allows you to scale down an image and get a much more accurate representation of both the luminosity and color of each actual pixel in your final image.
I just don't see this in practice. I wish I did, it would mean I could justify buying a 5Ds to produce better quality images. Theoretically of course you are right, the bayer array means that you need many more pixels on target to define colours, especially at the edges of things, but I just don't see it in practice.
The problem is all to do with scale; because of the resolution of even a 12 mp FF camera there is enough RGB coverage to define the detail we can see in say an A3 size print. The increase in colour definition that you are gaining from say a 5Ds is just lost in reducing the detail size down to an A3 print. That detail has just gone. Simple. If it hasn't then it was large enough for the lower mp sensor to define anyway.
I have found that the greater magnification of a larger format gives better tone and detail, so for instance again if we refer to a fairly common "large" A2 print (23.5 x 16.5 inches) a vertical three frame stitch from an old 12.7 mp 5D at about 24 mp is better printed definition than a single frame from a 24 mp M3. However as you reduce the print size, once again there is no difference.
Of course cropping is another thing, you are definitely limited in producing large prints from a heavily cropped 12 mp sensor
The one area where it really is a no brainer is cropping. Heavily crop a 6D image and definition & detail suffer but the 5DS if exposed correctly and sharply focused really stands up well and its apparent for all to see easily.
jayt567 said:
unfocused said:
Lara & Hugo de Chassiron-326 by Tom Scott, on Flickr
Lara & Hugo de Chassiron-335 by Tom Scott, on Flickr
N0. 78 1965 2.0 Porsche 911 Classic GT Cars (pre '66) Silverstone Classics 2016 by Tom Scott, on Flickr
Talys said:
Talys said:
Talys said:
