Help please! DNG and TIFF file size.

[garys37512407]

Hi there,

When I import my images to Lightroom Classic I automatically convert them to DNG's. The average size of a file is 50MB.

I apply any adjustments in Lightroom Classic and if the image needs any retouching I'll select 'edit in Photoshop'.

When I save the image in Photoshop it will get saved as a TIFF and will appear in the Lightroom filmstrip.

The TIFF images are now a huge average filesize of 250MB. I've even seen one as big as 900MB!

The size of the TIFFS is absolutly killing my storage space and I was wondering if their is a more efficient method.

Any help would be very much appreciated. I've been working this way for years!

Gary

[DS256]

You may want to try playing around with a few things:

• Review the size of your DNG after import to the original RAW file. Personally, I just leave my photos in their native RAW. When I do convert a NEF to DNG, they are about the same size. I've never seen any advantage of converting to DNG.
• I just had a quick look, and  a 21 MB, 6000x4000 Nikon file (NEF), going through PS created a 291MB TIF file. My LrC Preferences for PS are
  • File Formar = TIFF
  • Color Space = ProPhoto RGB
  • Bit Depth = 16 bits
  • Compression = None
• If you are concerned about disk space, look at using Compression. Personally, disk space is cheap so I don't worry but then I don't do much in PS.
• Others can comment on if there are advantages/disadvantages of using PSD rather than TIF

So, I don't think your sizes are abnormal.

[richardplondon]

There is no way round it, a TIFF is expected to save larger than a Raw file regardless whether that is still in the camera proprietary format, or in a DNG converted format. Even when no additional layers or layer masks are included in the TIFF, and even when not made as a smart object (which adds considerable file size penalty).

The Raw data - except in the case of some more unusual cameras - consists of a single number of say 14 binary bits, for each "pixel" location in the eventual image. That reports either a red-filtered, or a green-filtered, or a blue-filtered value.

Then after conversion / rendering to an RGB viewable image format such as TIFF, at minimum a separate red value and a green value and a blue value must be stored to describe this same "pixel". If editing to 8-bit format then this in crude terms totals 24 bits per pixel instead of the Raw file's 14 bits. If editing to 16-bit format then this in crude terms totals 48 bits per pixel instead of the Raw file's 14 bits.

Then additional pixel layers consume file size at the same rate. More so once layer masks / transparency are considered.

Selecting a lossless compressed format for saving the TIFF will certainly help. But there is no way IMO that anything can ever match the efficiency of (for those images where it is feasible) completing postprocessing within Lightroom Classic alone. The Raw file / DNG file is being stored in any case. Only once, no matter how many virtual-copy versions you may generate in LrC.

And LrC can go straight through to export / print just applying those parametric adjustments on the fly. An intermediate TIFF is only essential for editing tasks with no practical equivalent available inside LrC. This is an important difference from the more traditional workflow via ACR, whereby every image must go into PS before controllable output from that really becomes possible.

[Jao vdL]

Bottom line is. Try to do everything in Lightroom. As soon as you go into ANY external editor your filesize balloons because of the reasons @richardplondon lists. This is definitely possible for almost all images. Except if you do many really complex manipulations (i.e actually replacing image elements) you really don't need Photoshop even if you think you do.

[richardplondon]

This is all completely unsought comment, but I'll just put out there:

Lr (or ACR, or other parametric editor) is particularly good and efficient at changing what the image looks like.

Photoshop (or other equivalent application) is particularly good at changing what an image's content consists of.

So when using the two together, we have both changes to appearance and (sometimes) changes to content to consider. LrC's ability to be flexible and efficient at exporing variations in appearance and presentational usage, is to some extent negated by "freezing" an image into a bitmap file so it can be externally edited. But by deferring some parametric adjustments steps to only apply after an image returns from a round-trip into PS, some of that flexibility can be recovered.

That suggests a four stage approach where PS is involved (reducing to three stages where PS is not involved):

• initial Raw based processing limited to the best possible extraction of good quality general-purpose detailed picture content from the source - no cropping or straightening or 'aesthetics' as yet
• modifying in PS as necessary, of the full image, in 16-bit full colour, but still with no regard to presentational aspects - only concentrating on picture content
• more work in LrC to flexibly explore as many different "looks", colour and/or B&W treatments, crops and perspective transforms, soft-proofed output usages etc as your imagination permits - using as many virtual copies as you like - all referring back to that underlying single PS edit file and self-updating whenever you've gone back in to do some more content alterations there; or if you haven't been to PS, still referring back to the Raw. This presumes you did a good enough job of e.g. highlight recovery and artefact-free NR and sharpening in the first instance, that having / lacking direct access to the Raw data now makes comparatively little practical difference.
• LrC applies output-specific resampling, added sharpening, colourspace / print settings, bitdepth, page layout etc on the fly and affecting only that output instance, hence without the image being worked on ever becoming dedicated to any one particular form of output.

[Lumigraphics]

Um, I couldn't use this workflow. I find it MUCH easier to make global appearance changes in Photoshop because of layers, blend modes, and advanced masking that Lightroom can't touch.

[Jao vdL]

There is very little global that you can't do as effective ands quicker in Lightroom as in Photoshop. It is just another way of working. If you really need photoshop, you are stuck with the large file sizes. Make sure you use zip compression for the image bits and at least RLE for the layers but fundamentally if you don't want to lose significant quality, the imaging model in Photoshop necessitates very large file sizes. Nothing you can do about that.

[Lumigraphics]

Unless you need layers. I use them extensively.

[Conrad C]

---

@Jao vdL wrote:

Make sure you use zip compression for the image bits and at least RLE for the layers but fundamentally if you don't want to lose significant quality, the imaging model in Photoshop necessitates very large file sizes. Nothing you can do about that.

---

It’s important to note that this isn’t even just a Photoshop thing. You’ll get those file sizes in any image editing application that saves standard TIFF. It’s simply because of the math after multiplying bits * channels * width in pixels * height in pixels, and then adding layers and such.

In other words, the common perception that TIFF is huge is actually the reverse when viewed in context. In fact, uncompressed TIFF is the natural file size of an image, and raw and JPEG are unnaturally small, because each has some kind of major compromise. JPEG files are much smaller because quality is thrown out. A raw file is much smaller because you cannot view or edit it as a picture if it’s still in a raw state (raw data from the sensor)*; you can only view and edit a raw image after it’s been expanded into 3 RGB channels…expanding it to its natural file size.

*The only reason we can “see” a raw file straight from a camera is that the camera attached an RGB preview to it, or raw editing software generated an RGB preview for it.

[Jao vdL]

Yeah completely true. Nothing to do with Photoshop per se but simply owing to the number of pixels in the image and the need for it to be full RGB at 16 bits per channel instead of Bayer mosaiced in a raw file and 12 or 14 bits per pixel. So in the tiff file, if you don't do any layers, you have 48 bits per pixel while in a raw file you typically only have 12. A factor of 4 difference just due to the file model difference! This easily doubles and triples with just a few layers added.

[JP Hess]

I have learned to do nearly all of my editing in LrC, turning to Photoshop on rare occasions anymore. The Lightroom set of tools has improved dramatically to the point that, for me, Photoshop is rarely needed anymore for editing. I also use Nikon cameras. I experience very little space savings by converting to DNG and prefer to stay with the native NEF files. Storage space isn't all that expensive. I have two additional internal hard drives and two external hard drives connected to my computer, so I always have plenty of hard drive space to store my images.

[gary_sc]

Hi Gary,

My workflow is very much like yours with one big exception: I have all of my images stored on an external drive. A 4 TB drive to be exact. A long time ago I realized I had more images than what was good to be on an internal HD so I opted for two things: one is to place all of my images on an external drive and (2), to back up that drive to a 2nd external drive. You do not say what your OS is but if it's a Mac, I strongly encourage you to check out Chronosync, it is the best backup software available for the Mac. If you're on a PC, I'm sure there will someone around to advise for that. [In addition I also use a cloud backup service incase "the house burns down" kind of thing.]

And as far as DNG goes, I've been doing that since, well, forever. I do not need to sell you on the value of that, nor do I feel it's necessary to say it's not necessary. Just enjoy what you're doing.

One thing that DS256 said that is worthy of note but needed to expand upon is bit-depth. A 24 bit image (color) is 3 times larger than a 8 bit image (grayscale) but a 16 bit image (grayscale) is 256 TIMES larger than an 8 bit image. And I'll let you do the math to figure out how much larger a 48 bit image (16 bit for each color) is over an 24 bit image (8 bits for each color). So yeah, your image size is just fine.

I would NOT go the JPG route (only 8/24 bit possible). Hard drives are cheap, your images are priceless.

Good luck!

[Conrad C]

---

@garys37512407 wrote:

The size of the TIFFS is absolutly killing my storage space and I was wondering if their is a more efficient method....I've been working this way for years!

---

Even though many of us have done it the same way for years, changes around the workflow have increased the file sizes anyway.

10 years ago, a pro camera might capture for example 18 megapixels at 14 bits per pixel, so the raw file coming out of that would be around 24MB. Today, a pro camera might capture 45 megapixels at 14 bits per pixel for a raw file size of about 50MB. That increase has serious implications when converted to TIFF.

10 years ago, someone might convert the 18 megapixel/24MB raw file to uncompressed TIFF and get a file of about 50MB, as a 3-channel (RGB) file at 8 bits per pixel. Today, it’s more common to convert it at 16 bits per pixel, and that results in a TIFF file of over 100MB. So working at 16bpc in Photoshop instead of 8bpc is one reason today’s TIFF sizes are bigger.

Now if you do that with a 45 megapixel/50MB raw file from a current camera, converting at 16 bits per pixel, the 3-channel RGB file size, as uncompressed TIFF, is 260MB! Because two things have been compounded over the last 10 years: double the bits per RGB pixel, and 2x to 3x the typical number of megapixels per frame.

When you edit in Photoshop, you introduce the potential for bigger files beyond the above. Did you add a layer? Double (or more) the file size again, because each layer is like storing another image in the file. Add to that the Preserve Compatibility layer (flattened composite) which is required to see layered Photoshop files in other applications such as Lightroom, and the file size goes up some more. And file sizes go up faster at 16bpc than at 8bpc.

What can you do? Consider some tradeoffs, whichever are acceptable for your work:

• Use a camera that captures just enough pixels for your uses, not more than you need. Or set the camera to capture fewer pixels (e.g. mRaw, sRaw) when you don’t need them all. 
• Only go to Photoshop when you need to do something Lightroom can’t do. 
• Add only necessary layers in Photoshop, not an excessive number of them. 
• Set the TIFF file for pixel dimensions and bit depth no higher than required to produce the quality you need for final delivery.
• Use the LZW or ZIP compression option for TIFF. 
• Buy more storage. 

My solution was to buy bigger hard drives. But that is a problem now because the price of storage and other upgrades have gone up a lot recently, due to component shortages and other factors.

[garys37512407]

Thanks for taking the time to reply. There's been some really useful information to consider.

I've got to stick with my workflow as LR won't do everything I need. I'll continue in using external hard drives.

One other option would be to do all I can in LR, export the high res jpg, open and retouch that jpg in PS and save at maximum quality. This results in a filesize of around 40MB (I shoot with a Nikon D850). However, I'm not sure what effect this has on quality. There doesn't seem to be any lack of quality to the naked eye. The workflow isn't as smooth though.

Have a great day everyone and thanks for you advice.

[gary_sc]

There are two things for to consider about using JPGs. First is that they can only be 8-bit. That means that even saving at 12, you still may very well get posterization across a (say) sky when printing. The second is the issue of jpepegging a jpeg: simply every time you save a jpg, you are tossing out data.

So my question to you is the size of the image so important that your willing to potentially sacrifice the quality of your image? Considering the cost of hard drives, do you want to to be saving your images down a one-way street? By this I'm just trying to point out that once you go down the JPG road, any adjustments or corrections will be with a lower quality image format that you cannot reverse. 

good luck with your choice!

[Conrad C]

---

@garys37512407 wrote:

One other option would be to do all I can in LR, export the high res jpg, open and retouch that jpg in PS and save at maximum quality. This results in a filesize of around 40MB (I shoot with a Nikon D850). However, I'm not sure what effect this has on quality. There doesn't seem to be any lack of quality to the naked eye. The workflow isn't as smooth though.

---

If you come up with a workflow where you are happy with the final quality at the file size you want, then that is OK no matter what others say. It is true that at higher JPEG quality settings, there isn’t any quality loss to the naked eye. The quality loss can show up if that JPEG is manipulated further. For example, if you later realize you want to lighten the shadows a bit, JPEG will have a lot less room for adjustment before problems are visible, not helped by being limited to 8 bits per channel.

If saving storage space is such a high priority that 8bpc JPEG is being considered as an intermediate format, keep in mind that you might be able to maintain enough visible quality at less than the maximum setting. Above a certain setting (depending on the image content), file size goes up a lot more than the level of quality being preserved, in other words there’s a point of diminishing returns, as shown in these tests: An Analysis of Lightroom JPEG Export Quality Settings. That discussion is more about setting a JPEG level for final delivery, so for use as an intermediate format it would probably be better to use a compression setting closer to, but maybe not all the way to, maximum.

Also it sounds like you’re comfortable with a sort of one-and-done JPEG workflow through Photoshop. What I mean is opening as JPEG, editing, and saving as JPEG, so that there is no way to preserve edits on separate layers which you can do only with PSD/PSB/TIFF. For me, being able to change the edits later is important so I want to preserve those Photoshop layers, but you are free to give that up in exchange for very small files, as long as you get the quality you want. It’s all about the tradeoffs…