Software Review - Capture One Pro

Capture One Pro

Migration from Adobe Lightroom v6 to Phase 1 Capture One Pro v10

INTRODUCTION

By way of introduction, I have been using Lightroom to manage my image collection for a few years now, currently using LR 6 CS (not CC). For many reasons I am now looking for an alternative. While there are features I do like from all the other various offerings, such as LR, ex-Aperture, DXO Optics Pro, ACDSee and OnOne Photo Raw, plus the many DAM systems like Photo Supreme, Media One Pro (also by Phase One), Daminion, iMatch, etc, they are all missing something.

A good data management system should be implemented in a logical, clear manner, using easy to understand terms, with a simple and consistent layout, using standard criteria (such as scope, category, filter, group, sort) to define what images are selected, and all in a quick and robust query-oriented database workflow. In terms of an image database, once one or more images are selected using the above, it is from there that images can have information/metadata amended easily, while then providing the second part of the show, allowing beautiful raw/image manipulation, optional external manipulation, and output control.

After a few days of playing with Capture One 10 v 10.0.0.0.225, I decided to make a few observations. These are only based on my initial findings, and due to time constraints, is not as detailed as I would like, but a start. And yes, it is absolutely biased towards my own needs and opinions!

I should say firstly, I understand that this is a new release and likely to have a few flaws and bugs, plus I also understand that my point of view and requirements will differ from others. I acknowledge that my observations are just that, not to start any wars, and not to say this way or that way is right. I just thought I would start making a list of items as I see them, with my setup, in order to both provide feedback to Phase One, and allowing other potential new users to see what I have been up against in researching a migration path to the new platform.

To add some scope to this report, I thought it useful to summarise my situation. The computer I am using is a reasonable spec: 4-core processor, 8GB ram, 2GB GPU, internal SATA RAID 0 SSD for operating system and primary data (such as image catalogues), and external HD for the image library. I am running Windows 7 Pro SP 1 64-bit, with all latest updates and drivers. My images, about 70,000 of them, are stored as H:\Image Library\2001, with additional year  folders 2002 up to 2016. Within each year folder are the image folders, each with a date and title (eg folders such as 2012-03-20 Spanish Yachts, 2012-07-20 4wd Kimberly Area, etc). And lastly, about me; I am familiar with software, have been a programmer, understand database fundamentals, and have a fairly well mapped out workflow and the necessary system capabilities that stem from the subsequent requirements of that workflow.

I have certain requirements for such an image management system, as follows: -

1. Import, with renaming, and back-up, able to handle RAW, jpeg, tiff,  and assorted video  files
2. Robust and well organised database cataloguing environment
3. RAW and other processing with high quality rendering, combined with round-trip third-party image processing integration
4. Export and output capabilities
5. Flexible User Interface

Lightroom has much of this, but item 5 is where for me it falls down, meaning one must 'work within the Adobe module-based flow'. Items 1 to 4 are in many cases very well done in LR, especially the cataloging capability and speed. However, the look of the graphical user interface (GUI), such as where it places panels etc, the very limited 'grey' palette look, and a few 'prosumer' gimmicks starting to creep in, detract from the whole feel of it. Plus, while the images that it renders are pretty good, I am still looking for that 'something' extra. Perhaps I just need a change.

According to 'the brochure description', Capture One (C1) caters well to all items in my list, so let us see what my initial impressions are. To clarify, C1 uses both a 'session' and 'catalogue' approach, with the former geared towards small collections of 'projects' and the latter towards full library management. As I need a managed image database, I am testing the catalogue side of things.

PART 1 - Importing Process

Importing from Lightroom Database

This was very slow, and at times, seemed to stall completely. This method caused memory usage to go very high, and likely due to successive forced 'end process' actions through Task Manager and due to software crashes, the database became corrupted a few times, usually meaning a fresh start when the database check/recovery tools failed to restore it properly. It was a bit of a frustrating mess, with many 'beach-ball/clock-ticker/waiting' curser displays for sometimes hours, wondering if it was actually doing anything. Literally, many days were lost. Often C1 exceeded physical memory requirements, where C1's memory demands caused over-intensive use of virtual memory, bringing my system to its knees.

Importing from Hard Drive Folders

This seemed more reliable, but with a disadvantage of not picking up all the information from my LR database. On the import dialogue, selecting 'Open selection when import starts' caused high memory usage and slow import, as there was some conflict in importing the images while also trying to generate previews. Importing worked much quicker by instead selecting 'Notify when done'. This meant that during import, new images were not displayed, and thus not generated, lessening software load. So then, after all 70,000 photos were imported this way, previews were then built in a second stage, by selecting folders, one at a time, only moving to the next folder once previews were finished. It took a few days. To try and still get some LR metadata installed, I did try importing some image files along with .xmp sidecar files from LR, but that too seemed to cause slowness, and in one case, database corruption that I could not fix. The best way, was a fresh start, importing images first, generating previews second, with an idea to later adding metadata manually. Not ideal at all.

Renaming

The renaming side of things is a bit clunky, but I managed to get it to rename my files and put the folders in the right place during import. I tried both from a 'copy from memory card' and also a 'catalogue in place' scenario. Both end up with referenced files on the external hard drive, no problem. The only thing I couldn't get to work as desired was to be able to produce an exact back-up copy of my primary image drive onto my back-up image drive, as it seemed unable to produce the same folder structure on the back-up drive, just lots of files in one folder.

Thumbnail and Preview Generation

Thumbnail and image previews of raw files (NEF, CR2, RAF in my case) seemed fairly quick to build, about one file every 2 seconds. Jpeg files  were quicker, about 1 per second. Tiff files were slower, up to half a minute, depending on size. I found that some Jpeg and Tiff files failed to import. Not sure why, but possibly they had been previously modified using LR's metadata update (which, for non-raw files, saves directly into the Tiff or Jpeg). In the case of the Tiff files, also the possibility of multiple layers may have caused issues. The best way around this was to export the Jpegs and Tiffs as new, flattened Tiff files using LR, losing all the layers unfortunately, but thus allowing the files to be imported okay. Movie files had some issues too, and I had to do some conversions of older .mov files to .mp4 format, which seemed to work better.

I found that generation of previews was not affected too much by the size of preview files selected in preferences, but later on, I noticed that smaller preview sizes of say 1440 were displayed much quicker than the 2880 files.
There seemed to be an initial thumbnail generation on import that later seemed to be modified once C1 applied extra initial import adjustments when new previews were rendered. This caused a delay when displaying images the first time. It was a little quicker on subsequent viewing.

An interesting thing was that, using a disk drive viewing app to show what files are being accessed during any sort of preview/thumbnail generation, showed that sometimes the reading of dozens of files was in progress at any one time, likely being the reason for increased memory and thread usage. Trying to read-in and scan image files, alongside database insertion, tag, metadata, keyword organisation, while also generating previews, while remaining responsive in the GUI was not well managed it seemed. Perhaps a more 'sequential' approach (eg LR first imports the files into the database, and only then starts generating previews when that is done) would work better.

PART 2 - Database

Library

Folder Display - When showing folders, while it is nice to be able to see the whole hierarchy, would be nice to instead be able to just show selected parent(s) of an imported image folder. And then, if I select a parent, all images from child folders can be shown in the thumbnail display. This is not possible in C1.

User Collections - Groups, Albums and Smart Albums, are the equivalent of LR's Collection Sets, Collections and Smart Collections, all acting on the complete database. Projects add an extra layer, in order to isolate the smart searches to only those images in the albums that are within the project. It takes a while to get your head around this.

Folder and Collection Scope - The biggest issue is that only one folder or one collection can be looked through at a time. In LR, as well as singular selection using a click on a folder or collection, one can select multiple folders by using the standard ctrl-click, shift-click and ctrl-shift-click. Additionally, each parent folder or collection set can be selected to show the images from all ALL child folders or albums. And finally, each folder level, all the way up through to the root of the disk drive or top level collection shows the number of images of all child folders contained within, right through the hierarchy. C1 only shows the number of images for a folder that contains images, and only allows one such folder to be selected at a time. LR is so much more flexible, in fact almost indispensable, and that would be nice to have in C1.

Filters

Filter Choice - In C1, individual items can be clicked to show that sub-selection of images. Ctrl-click also works in order to select more than one item. The filters are grouped in what seems to be a preset order that cannot be changed. It would be better if the filter order was user selectable.

Selection - Each filter follows a typical standard of using OR when selecting the various values in one filter setting (eg selecting both  1 and 3 star images means show image that have 1 OR 3 stars), and using AND between filters (eg 1 star AND colour tag of red). However, be nice to also have the standard shift and ctrl-shift key selections for quickly selecting multiple groups of values (eg quickly selecting all rows of 1, 2, 3 and 4 stars). In addition, there should be an ALL and NONE selection for each filter.

As will be discussed below, when a collection or folder that has 100's of images is selected, it takes sometimes minutes for the filters panel to fully populate as the system seems to sequentially read the metadata for each image file within the folder or collection. It actually makes one wonder in this case if a database is being used at all. The process is very slow, whereas as in LR, it is virtually instantaneous, even when 70,000 images are selected.

Metadata

There are of course many ways to display and adjust metadata such as tags, labels, descriptions, comments, vender, EXIF and IPTC values, and very importantly, keywords. In Capture One, the way to change metadata is not always consistent. There is a both 'Metadata' tool, 'Keyword Library' and 'Keyword' tool panels. The metadata panel seems to only show items that pertain to the selected image (the 'Primary Variant') even if there are many selected images. If one makes a change or entry to the field, it will only adjust the primary variant.  In contrast, LR shows all data from all selected images, and where a field is different across images, marks the field as such. Any change to a field will change all images selected, plus there is a copy and paste, and a sync method; you get to choose. In Capture One, this is achieved using a 'copy and paste' syncing method, which is not quite as flexible. In C1, it is possible to set some items, like ratings and colours, across all selected images using the pull-down Adjustments menu option without using the copy-paste method. A bit inconsistent.

Another difference is that in Capture One, one has a metadata panel to display data, but to filter images using that data, you have similar items displayed in the filter panel which can then be selected, whereas in LR, one simply has to press  a little arrow next to the metadata selected in the metadata panel, and all images with that value are displayed. In LR, there is also a separate filtering panel, which, because of its horizontal placement on top of the image area, is one part of LR I do not like.

Now, for the keywording part, that could take a whole discussion by itself. The way that it is done in C1 is not bad, but there are a few things missing, namely, no way to have 'private' keywords or keyword sets/groups (i.e. ones that only are useful for organisation and do not export with an output file). There is a way around this using 'shared' keyword lists, and then stipulating which of these lists are output, but not quite so elegant as being able to set the private keyword sets/groups individually in the main keyword library. Also allowing synonyms does not look possible, and lastly, no keyword defined lists for quickly setting topic-based default keywords en masse. The good thing is that in C1, the keywords can be hierarchical, and also, where a number of images are selected, keywords show up as being in 'some' of those images, and can be set across all the selected images, without needing to use the cut and paste method. LR is slightly better in many regards.

So, yes, a bit inconsistent all in all. An idea, why not use a combined 'filter/metadata/keyword adjustment panel'. The three essentially show similar information; it is all data. The current filter panel, shows each filter field, listing the many values of that field are currently used in the database, with a number to the right indicating how many images in the current collection/selection pertain to each item. Then, with that same panel, the item can be clicked on a plus next to it to assign to all selected images, or clicked on minus to clear all selected images of that item. Daminion Image Viewer uses a simple switch to change the filter panel mode from filter mode to assignment mode. It would also be nice to have an approach that allows the adjustment of metadata values across all 'selected' images rather than using the copy-paste method.

Performance

This is, for me, is the glaring issue with Capture One, especially when comparing to other database driven apps like LR. Speed of the typical database actions (filter, sort, group, search) is abysmal. As an example, when a smart album is clicked, the number of image thumbnails and associated filter data fields takes a while to build up, literally minutes when the number is more than say 5000 images before it has finished going through them all. You can see the count for things like total number of images, dates, filter items, thumbnails, etc, all slowly counting up from 1 until the total has been found. There is a little 'pie-slice' counter at the top of the display showing say '1 of 5000 images', and this slowly counts up over a matter of minutes, as it slowly goes through and redisplays thumbnails, eventually in the right order. Even though the thumbnails and images may have already been generated previously, these are displayed with quite a lag. I don't get that. Scrolling through a page of thumbnails, usually displays lots of blank boxes, until it slowly displays thumbnails in them, all very slowly. During the display of thumbnails, the processor kicks into very high. In LR, a similar operation is almost instant, and can display all my 70000 image thumbnails within a few seconds of loading the program. So, what is going on with C1?

The above process is quicker in C1 when the disk drive holding the referenced images in not attached to the compluter, but still very slow. When the disk drive full of images is attached, when a folder is clicked on, you can see that the thumbnails slowly get displayed, but as well as high processor usage, the disk drive of images is also being accessed, as though C1 is doing some comparison/reference between the database and image cache and the original image files. LR, on the other hand, only accesses the disk drive of original images if it needs access to the original files for a 100% display or editing. So, what is C1 doing with the images when just clicking on a folder or collection? One thing is for sure, it really slows down the display of thumbnails, metadata and images, and access to all other software functions is somewhat slowed by this process. The browsing capability is just too slow to be usable. It makes looking for a particular photo in amongst even several hundred really slow and drawn out. It really sucks.

From what I see, it looks like Capture One may in fact not be using a typical relational database query action for such things, but instead relies upon a flat-file sequential interrogation technique, which essential scans through tables of information, or even the files themselves, populating associated index fields as it goes. Maybe, maybe not. But a query to count say 5000 records searched from 70000 should be very quick. Even an Excel spreadsheet can group, search and sort such a group in just a couple of seconds. So, something is amiss. The insistence of referring to the original images when it should just be instantly pulling thumbnails from the cache and metadata from an indexed database does not make sense.

I also found that, certainly with my GPU, things were much slower and the program more likely to crash. So I set the preferences to not use the GPU. This made things faster and more robust, but not by much. It could be that the software relies on a modern, supported GPU, to be able to display things more quickly. Perhaps, but, since there is some delay caused by reading from the disk drive of images, a GPU will not fix that. That is software process-related. I still go back to thinking there is something fundamentally wrong about the way cached images are referenced and metadata is queried from the database.

There are conflicting reports on speed and performance on the internet. Some people are satisfied, while others, like myself, are not.  Without a doubt, this is to do with a variation in both set-up, hardware, and even possibly software conflicts.

• In terms of setup, if one uses sessions instead of the catalogue approach, perhaps only accessing a few hundred rather than tens of thousands of images at a time,  this will make a huge difference. I found performance absolutely tolerable when only 500 images were in the catalogue, but things went downhill the more I threw into it.
• In terms of hardware, yes, if one uses a Xeon processor, 64GB DD4 ECC 2400 ram, a Quadro 8GB GDDR5 PCIe3.0 GPU, and latest PCIe high speed SSD disk drives, I am very sure C1 would be somewhat faster, but everyone has a different level of hardware, and this will make a huge difference if the software is not efficient.
• And in terms of software conflicts, as has been seen in many places, even having non-updated GPU drivers can cause massive slowdowns, or even crashes. A software platform has to cope with a lot of variation, and stability and maximizing performance over this variation does take some work.

So, without pointing out every glaring detail of what is slow (since it is almost everything anyway), my recommendation is that the underlying database structure and actions need to be completely re-written for this software. I may be wrong, but based on what one can visualise on an 'average' computer hardware setup, it just does not appear that the system is using any sort of proper database querying at all. So, that, fundamentally, needs to change if it is to be a serious DAM product. If Phase 1 actually do not want it to go that way, then fair enough, but that will deter many from switching, regardless of how good the RAW output is.

PART 3 - RAW PROCESSING

I have not played around much with this side of it, but so far I find most of the tools comparable, sometimes better, sometimes lacking than say LR. I leave the review of this side of it until a later date.

PART 4 - EXPORTING

This seems to be very flexible, maybe not quite as simple as say LR, but certainly on par with other systems in terms of capabilities. Exporting in a number of formats at one time is in fact quite powerful, if somewhat more difficult to setup. I have no major problem with the functionality provided. I also leave a review of this side of things for a later date.

PART 5 - GUI

Toolbars - The design is fairly flexible, although there are a few quirks and some suggestions. Very good to be able to totally configures the toolbar and its placement. Would like to be able to have both a right and left docking toolbar, or better still, as many multi-tabbed toolbars as I need. Screens and resolution are big these days, so multiple docked toolbars would really be able to make use of the real-estate. The way around this right was to have the built in toolbar on the left and use floating panels over on the right. Functional but not so elegant.

Keyboard and Mouse - The ability to change keyboard shortcuts is, for me, one of the best parts of the whole thing. However, mouse implementation is not always how 'I like it'. Be nice to be able to decide what the scroll wheel action is in the viewer and browse windows, as in zoom in/out, move to next/previous image or scroll n-images at a time. Either by selecting the functions in preferences, or providing all by following a 'sort of' standard where the scroll wheel by itself moves to next/previous image, ctrl-scroll adjust zoom, and alt-scroll moves a few at a time.

There are lots of small issues, such as it would be nice if the size of the Activities windows was adjustable, the ability to stack ANY files together rather than just variants of a master image, and it would be good to have some sort of scaling and colour adjustment on the GUI fonts. But all these smaller issues could be lived with for now.

On the whole, I prefer the layout of C1 to LR. Working with one display, thumbnails always in view, without having to change 'modes' like in LR, feels better for my way of thinking. And either way, work-spaces in C1 can be used if one prefers a modular approach, so tons of flexibility really.

SUMMARY

So, out of my original five requirements of a good image management system, what is missing from Capture One?

Capture One Pro has the potential for being brilliant. It already coaxes me with it's cleaner, more flexible graphical user layout, excellent adjustment tools, great rendered output, and most of the facilities needed for a good DAM. But it is within this last area where the Achilles Heal is found. The one drawback that cripples the overall package, is that the 'implementation' and 'performance' of the database and caching systems is poor. It is slow, seemingly inefficient in memory, processor and disk management, and ultimately hinders one from using the rest of the package to its full potential.

What it needs is a real thorough going-over in the database and cache area. I imagine that will not be simple. We are on version 10, and I am sure that much of the code has become somewhat 'baked-in', instead of modular. However, I believe that it is possible to re-write the DAM portion, with robust and efficient database access, and then call the new modules in all the places where they baked in the old database functionality. Yes, it would take work, but it would then propel C1 to the very top, and provide an excellent option for those that feel abandoned by Apple's Aperture and those becoming more and more dissatisfied with LR.

I hope Phase One give it a shot.

Cass - Photographer.

Software Review