Can Using Tiff Instead of Jpg Files Overcome Resolution Issues?

[David B]

Hi all. I have been pondering whether to buy the R5 or the R6, and after having been loaned an R6 and several lenses for a few days by my local Canon dealer I've pretty much decided on the R6. The only niggle is that 20mp sensor. Most of my work goes in magazines and websites and resolution isn't an issue, but there may be the occasional shoot that needs to be blown up. What I want to know is this: a 60mb RAW file off a 5DMkIV gives Tiff and Jpg files of approximately 80mb and 20mb respectively. It's more or less the same off an EOS R. On the R6, the RAW file is about 24mb, and Tiff and Jpg are 50mb and 10mb respectively. So if I were to take large prints off the R6's Tiff files, will this make up for the relatively lower resolution RAW files? TIA.

#R6 #resolutiontalk #TiffFiles

 

[Jake Shoots Birds]

Before moving to Canon mirrorless I spent 6 years shooting wildlife with a pair of 20MP Nikon D500s. I own an R6 as well as an R5 and that are my primary shooters.

File size does not equal resolution. A fully processed TIFF file will have no better resolution than a JPEG of the same dimensions, it will just be bigger (and provide a better platform for future edits, which is why I say "fully processed"). Want bigger then you'll want to turn to one of the resize programs. I personally used Topaz Gigapixel AI. Is it good? I have taken a 4000x2250px image shot with an Autel EVO drone (original model) that was cropped down to 1600px on the long end and produced a high quality 30"x45" canvas print for a customer.

The megapixel discussion is one I refuse to enter because of the AI I've used to avoid ever need a camera that shoots more than 20-something MPs. My reason for the R5 and R7 is primarily because I need crop resolution and not full image resolution. If I crop down to 20MP's I'm thrilled.

 

[Kwazy]

That's an interesting take, Jake.
You feel that the Gigapixel is a good-enough substitute for higher Megapixelage?

I've had pleasant results with Topaz's upscaling on portraits (was pretty cool to take some 2 megapixel shots from 1997 and upscale them to normal resolutions), but anything it's not specifically trained on (like macro in general or insects in particular) it doesn't do a very good job.

 

[Jake Shoots Birds]

That's an interesting take, Jake.
You feel that the Gigapixel is a good-enough substitute for higher Megapixelage?

I've had pleasant results with Topaz's upscaling on portraits (was pretty cool to take some 2 megapixel shots from 1997 and upscale them to normal resolutions), but anything it's not specifically trained on (like macro in general or insects in particular) it doesn't do a very good job.

I feel that if you are satisfied with the detail you can capture in a lower megapixel image then buying a more expensive camera just for the MP's in completely unnecessary.

As a wildlife photographer I'm used to the phrase "pixels per bird" when speaking of resolution. Are the pixels per bird sufficient to render all the necessary details in that bird? A 24MP full frame camera is not going to give you the same number of pixels per bird as a 24MP cropped sensor camera with the same lens because you've got added reach depending on the crop factor. You're also dealing with other factors in a cropped sensor that impact image quality.

But, if you are able to render an image of sufficient detail and image quality on a sensor of any size that you could present it on Social Media at 1920px on the long side with no visible anomalies then I would posit that capturing that same image with a more densely packed sensor will likely not improve the quality to the extent that you would notice a difference between the higher MP image and an equally sized one using AI software to upsize it.

In other words, it will never be a substitute in the field when taking the photo from a pixels-per-subject/need-to-render-small-details point of view. But if you took the same well-exposed image with an R6 and an R5 and I up-sized the R6 to 45MP's and then generated both social-media ready images as well as large prints you would be hard pressed to tell the difference. Could you pixel peep in Photoshop and see it? It would depend on the subject, but sure, I'm thinking you would. But not so much in 95% of real world applications of most photographs - and nearly any social media context.

 

[Kwazy]

Ok, so you're not talking about using Gigapixel to take a 20MP photo and blow it up to billboard size.
But then I'm wondering, what's the point of upscaling if you've already captured all the detail you wanted for an instagram post? Does it create that coveted crispness that some of the better photographers' work seems to possess?

 

[David B]

Before moving to Canon mirrorless I spent 6 years shooting wildlife with a pair of 20MP Nikon D500s. I own an R6 as well as an R5 and that are my primary shooters.

File size does not equal resolution. A fully processed TIFF file will have no better resolution than a JPEG of the same dimensions, it will just be bigger (and provide a better platform for future edits, which is why I say "fully processed"). Want bigger then you'll want to turn to one of the resize programs. I personally used Topaz Gigapixel AI. Is it good? I have taken a 4000x2250px image shot with an Autel EVO drone (original model) that was cropped down to 1600px on the long end and produced a high quality 30"x45" canvas print for a customer.

The megapixel discussion is one I refuse to enter because of the AI I've used to avoid ever need a camera that shoots more than 20-something MPs. My reason for the R5 and R7 is primarily because I need crop resolution and not full image resolution. If I crop down to 20MP's I'm thrilled.

Thanks for clarifying that, Jake. So you're saying there's no point at all in saving an image as a Tiff if all processing is complete? A Jpg file sent to the printer will be no worse than the same image in Tiff format?

Personally, I've had 18"x22" prints made for framing off an 18mp Canon 600D (and one or two from even smaller sensors) , and they looked fine, but that was 10 years ago, and I feel large prints would be expected to have better quality today.

 

[Kwazy]

A Jpg file sent to the printer will be no worse than the same image in Tiff format?

Keep in mind, that when saving a file as JPG after any processing you do on your computer you can save it with various amounts of compression (e.g. default is usually 90%). This may affect quality. If you want the best quality possible and are starting with a TIFF file, I'd stick with the TIFF if your printer accepts that. But all this is not to say that file size per se has anything to do with quality.

P.S. Personally, I only deal with JPG files, but I do crank up the quality to max available. 12 in photoshop, 100% in other programs. Works well enough.

 

[Paul Iddon]

I save my files as a TIFF, then reduce the size and save as a .png before the jpg file last.

Paul.

 

[Jake Shoots Birds]

Thanks for clarifying that, Jake. So you're saying there's no point at all in saving an image as a Tiff if all processing is complete? A Jpg file sent to the printer will be no worse than the same image in Tiff format?

Personally, I've had 18"x22" prints made for framing off an 18mp Canon 600D (and one or two from even smaller sensors) , and they looked fine, but that was 10 years ago, and I feel large prints would be expected to have better quality today.

The main thing is to understand the internal resolution of the file. You want 150-300ppi on the image you send to the printer (this measurement ONLY matters when printing, not when posting online. Whether that's JPEG or TIFF matters not. I always us Gigapixel AI to upsize to the precise print size (in inches) at a resolution over 150ppi and then save it as a JPEG to send off.

 

[David B]

OK, so I just downloaded a trial version of Gigapixel AI and just ran it mostly auto on one of my 20mb jpgs off the R6, and it banged it up by 4x, changing my 300 dpi 41cmx28cm image into a 166cmx113cm at 300 dpi. Now I didn't fiddle with the face recovery and grain reduction and other controls. I do notice some odd blurring on parts of the faces, etc. I take it all of this can be sorted out by tweaking the settings? Image 1 below is he original jpg, 2 is a 100% crop from it, and 3 is the same area on the Gigapixel image, showing the odd blur.

 

[Kwazy]

I've had the same issues with this software, where it randomly decides to smooth some part of a person. There's a threshold slider that lets you adjust some settings, but they don't really give the user as much control as I'd like.

Great photo though. Would love to see more of your work.

Edit: On a second thought, are you using the Gigapixel, or their combined PhotoAI? I don't think Gigapixel should be smoothing things. It's some "photo enhance" feature that's doing that.

 

[Copterdoc]

OK, so I just downloaded a trial version of Gigapixel AI and just ran it mostly auto on one of my 20mb jpgs off the R6, and it banged it up by 4x, changing my 300 dpi 41cmx28cm image into a 166cmx113cm at 300 dpi. Now I didn't fiddle with the face recovery and grain reduction and other controls. I do notice some odd blurring on parts of the faces, etc. I take it all of this can be sorted out by tweaking the settings? Image 1 below is he original jpg, 2 is a 100% crop from it, and 3 is the same area on the Gigapixel image, showing the odd blur.

Just a guess but looking at the face and the focus, this seems like a bit of a depth of field issue and not a resolution issue. Also, Gigapixel AI also has a Sharpen AI-associated program that may help some of this. Also, they have a newer version of a combined program that may provide all of the Sharpen Resize, and other functions in it. I'm looking into this now.

 

[David B]

It's the Gigapixel AI

I've had the same issues with this software, where it randomly decides to smooth some part of a person. There's a threshold slider that lets you adjust some settings, but they don't really give the user as much control as I'd like.

Great photo though. Would love to see more of your work.

Edit: On a second thought, are you using the Gigapixel, or their combined PhotoAI? I don't think Gigapixel should be smoothing things. It's some "photo enhance" feature that's doing that.

It's the Gigapixel AI image upscaler that I'm trying out: https://www.topazlabs.com/gigapixel-ai I guess I'll have to do some more with it and see if there's any way of controlling it better.

Thanks for the compliment, Kwazy. You can see my work on insta: @sonofthemorninglight or https://www.instagram.com/sonofthemorninglight/ Some of the most recent shots are with the R6 I'm trialing.

 

[Kwazy]

You can see my work on insta: @sonofthemorninglight

Followed you. Keep it up, superb work.

 

[David B]

Followed you. Keep it up, superb work.

Thanks, mate. Love the bugs btw.

 

[David B]

Just a guess but looking at the face and the focus, this seems like a bit of a depth of field issue and not a resolution issue. Also, Gigapixel AI also has a Sharpen AI-associated program that may help some of this. Also, they have a newer version of a combined program that may provide all of the Sharpen Resize, and other functions in it. I'm looking into this now.

Well, I'm trying to use it just for upscaling so maybe I should turn everything down to the minimum?

 

[David B]

So I tried it again, with everything switched off (see screenshot) and there doesn't seem to be any odd smoothing or blurring.

 

[David B]

So I just ran the CR3 file of the above image through Gigapixel at x2 and it gave me a DNG file that can be fully processed in CR (it's huge, though at 500mb), and the 92cmx63cm 300dpi Jpg off that came in at a 50mb file. I think this will solve my resolution issues.

 

[David B]

Sorry about the chain of comments, guys, but I'm kinda excited to find out that the newer versions of CR have this 'Super Resolution' feature that doubles the size of the CR files, but still manages to give an output jpg file that's a more manageable 38mb. It's not so versatile as Gigapixel, but it is free, and plugs right into my current workflow with CR + PS, without adding in a standalone $100 piece of software. And since I'm mostly interested in the scaling feature, I may go with this.

 

[Kwazy]

Did you notice the colorful artifacts in the screenshot you provided? Those might show up when printed! Looks like a glitch in the software.

 

[David B]

Did you notice the colorful artifacts in the screenshot you provided? Those might show up when printed! Looks like a glitch in the software.

Yeah I think it depends a bit on the order of the steps. What one does before and after the 'enhance' command. I had another go and there aren't any artefacts now. I'll have to just play around with it a bit and see if the results are satisfactory each time. This is a 200% crop.

 

[Jake Shoots Birds]

Did you notice the colorful artifacts in the screenshot you provided? Those might show up when printed! Looks like a glitch in the software.

Or in the use. It's set to low resolution and a 4x increase with no noise suppression was done in a single pass. Nothing is magic. Sometimes it takes a bit of work to understand how to best upsize something. What works for one will not work for another.

 

[Kwazy]

Or in the use.

That's certainly possible.
I was going off a previous experience. I had a somewhat similar-looking but larger artifact in an image produced by DeNoise that went away after I re-processed it with the same exact settings. I contacted the support about it, and the rep confirmed that it's a glitch and the workaround is to reprocess. : /

 

[Jake Shoots Birds]

That's certainly possible.
I was going off a previous experience. I had a somewhat similar-looking but larger artifact in an image produced by DeNoise that went away after I re-processed it with the same exact settings. I contacted the support about it, and the rep confirmed that it's a glitch and the workaround is to reprocess. : /

Yeah, I'm not saying I don't find artifacts in the topaz stuff, and Denoise in particular. I'm just saying that the presence of artifacts can sometimes be mitigated based on settings. And sometimes not.

 

[ctitanic]

Hi all. I have been pondering whether to buy the R5 or the R6, and after having been loaned an R6 and several lenses for a few days by my local Canon dealer I've pretty much decided on the R6. The only niggle is that 20mp sensor. Most of my work goes in magazines and websites and resolution isn't an issue, but there may be the occasional shoot that needs to be blown up. What I want to know is this: a 60mb RAW file off a 5DMkIV gives Tiff and Jpg files of approximately 80mb and 20mb respectively. It's more or less the same off an EOS R. On the R6, the RAW file is about 24mb, and Tiff and Jpg are 50mb and 10mb respectively. So if I were to take large prints off the R6's Tiff files, will this make up for the relatively lower resolution RAW files? TIA.

#R6 #resolutiontalk #TiffFiles

A cRAW of 20 Mb can end in a Tiff of 80 Gib. It's all about compression method used in the file format. Nothing to do with resolution. Both files have the same resolution. I do not think that your work will be affected in anyway by the level of compression. It's the same as should you use RAW or cRAW. The technical explanation will tell you that by using cRAW you will end information because of the compression. The reality is that you won't be able to notice the difference. Compression has changed a lot over the years and the loss of information compression is minimal nowadays.

Now, a 20 Mp sensor has lower resolution than a 24 Mp sensor. But there are other parameters that go into image quality such as dynamic range. You can have a 20 Mp camera giving you better images than a 24 Mp camera. Mp is not all. Because the R6 is a newer camera than your 5D Mk IV I would say that you wont be affected by the 4 Mp difference.

 

[ctitanic]

Regarding upscaling. I use Corel Paintshop. I would recommend to use denoise or topaz sharpen before applying any resizing software. That change in the editing workflow will help you a lot.

Many people use just DeNoise. I mostly use Topaz Sharpen. The combination of both is very useful too.

 

[David B]

Thanks for all the useful tips and advice, guys. In the end I decided to buy the R6, and a second lens with the difference in price to the R5. So got myself the RF24-70/2.8L and the RF14-35/4L. What I think tipped the scales finally towards the R6 was the fact that I needed to fork out for that highspeed CF express card to be able to have a backup card, even though I shoot very little video.

I will carry on with the upscaling feature in CR for now, if needed, and consider buying GigaPixel if if I have to.