That’s a lot of work over and over, for each edit adjustment.
In rawproc, each tool has a copy of the internal image, copied from the previous tool’s result. I’ve tried to enforce the discipline that, when a particular tool is changed, the pipeline is re-processed from that tool to the tool marked for display. Even then, heaven help me if there’s a denoise in the way…
Thing is,there’s a price to be paid for edit-by-edit viewing resolution. Me, I pay by chewing my way through memory at an alarming rate…
@europlatus the articles for each tool will state how accurate the preview is, or whether the tool has no effect at all at less than 100% zoom. In fact, that is the very first paragraph of each relevant article. e.g. see Noise Reduction and compare it to Tone Mapping.
Yep, I referred to that in the last part of my earlier comment and it led me to query whether what XavAL wrote earlier in this thread is accurate and whether OP had misunderstood. You can’t see the effect of some tools at less than 100%.
I think OP was under the impression that effects could be seen at any zoom level, although they could be inaccurate:
Ok. Let me rephrase: when you use a tool labeled «1:1» you will only see the real, actual, exact effect at 100% zoom. That’s what every programmer is telling you. That’s what I didn’t say clearly in my previous post.
Now let’s see it from a practical point of view. Well, let me rephrase it in other words: from the experience I have in a few «1:1» tools (including the wavelets tool).
If you use a tool that modifies the image at pixel level (that is, it is capable of modifying single pixels to give some useful change to the image), and you tweak the settings so the change is clearly visible at 100%, then you will be able to see an average of the change when zoomed out.
You won’t see the real, precise effect, but you won’t see all and every pixels either. So you will be able to see what is going on. E.g.: if you change the contrast, you will be able to see if you like the overall effect, to go then at 100% and fine tune it. On the contrary, if you have gone too far with the settings, you will clearly see it zoomed out.
It happened to me with noise reduction, defringe, wavelets, sharpening, microcontrast, haze removal, … But you have to know each tool to learn to see what it is doing (that means, you won’t master the tools on the first image you process). And it will depend on the settings of the tool: if you make a tiny modification on the tool, don’t expect to see any change while zoomed out.
That’s what Filmulator does, except at two resolutions.
The full-res preview only actually gets processed when you stop moving the sliders long enough: it gets preempted by the quick preview, and it only runs the steps that are after what changed.
Thanks XavAL, although Rawpedia says that preview effects are “disabled” for some tools at zoom levels below 100%.
I completely understand what you are saying, and I believe this is how most imaging software works (i.e. you don’t get accurate rendering at less than 100%, but you will see some effect if you’ve made significant changes).
But RT seems to work differently in that you won’t be able to see the effects of some tools at less than 100% because they are actively disabled.
I use darktable more often but have recently become more interested in RT and want to know how it works. So, as I’m quite new to using RT, I can’t actually comment much on my own experience. I’m just repeating what Rawpedia says and that we shouldn’t be able to see any preview effects for some tools below 100% zoom.
So, in your experience, have you seen the effect of e.g. noise reduction/sharpening at less than 100%?
No I have not. In fact I think that generally my files are sharp enough and I don’t like too much sharpness. Right now, I have CS (or RL in ART) + RL sharpening at default settings in Resize module, and I’m happy with that.
I mean, it’s easy enough to just try on my own whether a tool seems to change the zoomed view. Some tools do, some tools don’t. That’s fine.
You all definitely increased my understanding of the rationale behind these concepts, how to work around their limitations, and where to find out more about them.
So thank you all very much, this has been enlightening.
I do wonder about the reasons for one program to be fast where a similar program is slow. Like, what corners needed to be cut and what limitations imposed to get Capture One as fast as it is; and what are the architectural differences that make Darktable slower, and RawTherapee use its 1:1 trick. Such an investigation of RAW processor architecture would be super interesting (but is obviously not going to happen for our closed-source peers).
In the end all my explanations only work with the wavelets tool.