James Ritson

Hi @smadell, hopefully this will give you a clearer answer! What upgraded GPU option are you considering over the standard one, and is it the 21" or the 27" model you're after? Looking at the specs, for the 21" the cheapest model has Intel Iris Plus graphics whereas you can upgrade to a Radeon 555X or 560X. With the 27" model the cheapest option is a 570X, and the most expensive upgrade gets you a 580X. Any of the upgraded GPU options will give you an appreciable boost in performance with pretty much all operations in Photo: the vast majority of raster operations are accelerated with a few exceptions—for example, blend ranges will currently fall back to software (CPU). Vector layers/objects are also not accelerated. The biggest advantage I've noticed is being able to stack multiple live filters without significant slowdown: I typically use unsharp mask, clarity, noise reduction, gaussian blur, motion blur and procedural texture filters with compositions and photographs, and being able to use compute is incredibly helpful as it keeps the editing process smooth and quick. Export times are also drastically reduced when you have live filters in your document: in software (CPU), as soon as you start stacking several live filters the export time can easily take over a minute, whereas with compute on the GPU this is reduced to no more than several seconds. However, the biggest limiting factor in my experience has been VRAM, and you will need to scale your requirements (and expectations) in accordance with a) the screen resolution, b) the pixel resolutions you typically work with and c) the bit depth you typically work in. To give you a rough idea, 4GB of VRAM is just about sufficient to initially develop a 24 megapixel RAW file and then work in 16-bit per channel precision on a 5K display (5120x2880). If you move down to 4K (3840x2160) 4GB becomes a much more viable option. This is somewhat subjective, but I would say forget about 2GB VRAM or lower if those are your baseline requirements—you simply won't have a good editing experience as the VRAM will easily max out and swap memory will be used, incurring a huge performance penalty. Ironically, if you can't stretch budget-wise to a GPU with 4GB or even 8GB of VRAM, the Intel Iris Plus graphics option may provide a better experience since it dynamically allocates its VRAM from main memory, therefore it can grow to accommodate larger memory requirements. From early Metal Compute testing I often found that disabling my MacBook's discrete GPU (with 2GB VRAM) and only using the Intel integrated graphics would alleviate the memory bottleneck. I believe the memory management has improved since then, but if you're on a budget that is an option to consider. However, if you're looking at the more expensive iMac models, I think you should weigh up your requirements and what content you work with in Photo. Here are a few scenarios I can think of: 4K resolution, light editing—development of RAW files, then adding some adjustment layers and live filter layers—you could get away with a 2GB VRAM GPU. 5K resolution, light editing—definitely go with a 4GB VRAM GPU. 4K resolution, moderate editing—development of RAW files, lots of adjustment layers and live filter layers, some compositing work with multiple layers—go with a 4GB VRAM GPU. 5K resolution, moderate editing—4GB VRAM GPU minimum. 4K/5K resolution, heavy editing—working with compositions that have many image/pixel layers, clipped adjustments, live filter layers—absolutely consider an 8GB VRAM GPU. 4K/5K resolution, 32-bit/16-bit compositing e.g. 3D render work, using render passes, editing compositions in 16-bit—8GB VRAM GPU minimum. However, if budget allows, do also consider the possibility of an external GPU: this might even work out cheaper than having to choose an upgraded iMac model. In the office I have a MacBook with a 560X GPU that has 4GB VRAM—this is sufficient for demoing/tutorials at 4K or the MacBook panel's resolution (3360x2100) but I work with an external monitor at 5K and for that I use an eGPU enclosure with a Vega 64 that has 8GB VRAM. The additional compute power is incredibly useful, but it's mainly the larger pool of VRAM that helps out here. You don't have to use a Vega, I believe the new Navi cards are supported so you could look at a 5500XT with 8GB VRAM which is a reasonably cheap option (although you would still have to get the GPU enclosure...) As you mentioned, your timeframe is 6 months so it might be worth waiting for Apple to refresh the iMac lineup as they will hopefully switch to Navi-based cards like they have with the new MacBook range. No doubt the cheapest option will have 4GB VRAM but models with 8GB should also be available. Apologies for the wall of text, but hopefully that gives you some more information to work with!

Hi Sean, best advice is not to touch the default colour settings in Preferences: leave the RGB/RGB32 colour profiles set to sRGB. RAW files are not mapped to sRGB/Adobe RGB etc, the camera setting is used for the in-camera JPEG and may also be used to dictate the initial output profile with RAW development software. After demosaicing, the colours are translated from the camera's colour space to a working space, then mapped to the output colour space which you get to choose. By default this is sRGB. If you wish to use a wider profile, check the Profiles option in the Develop Persona and choose an option from the dropdown. If you want to edit in ProPhoto, you can pick ROMM RGB which is the equivalent that ships with the Affinity apps. As mentioned above, the easiest option is to keep everything as sRGB. If you choose to work in a wider profile like Adobe RGB or ROMM RGB and you export to JPEG/TIFF etc you will generally want to convert to sRGB during export because other apps/web browsers/image hosts may not be colour space aware or colour managed. To do this, click More on the export dialog, and from the ICC profile dropdown choose sRGB IEC61966-2.1 There is a good argument for using a wider colour space if you tend to print your work, but you also need to be able to see these colours on your screen—do you know what coverage your monitor has of Adobe RGB/ProPhoto RGB, i.e. have you profiled it? If not, the safest tactic is simply to stick to sRGB again. Hope that helps!

Hi @GarryP, the new tutorials are very much focused on this approach, are you referring to them or the legacy set? For example, the new set (https://forum.affinity.serif.com/index.php?/topic/87161-official-affinity-photo-desktop-tutorials/) has a Selective Colour tutorial amongst other tutorials that look at specific filters and adjustments—Curves, Gradient map, Levels, Shadows/highlights, Displace, Denoise, Radial blur, Clarity, Channel mixer, White balance, Black & white, Zoom blur, Procedural Texture, etc... With the new tutorials, there's less of a focus on multiple techniques in one video, although sometimes this approach may be required. The videos also need to be kept fairly to-the-point because we now get them transcribed and translated into 8 different languages (which is particularly expensive)—there's also the issue of engagement, where many viewers don't make it through lengthy videos, and we have to take that into consideration as well. I would love to be able to produce more videos in-between major releases but time is quite limited, so it will tend to be during or just after updates to the app when you'll find more tutorials being made available. Hope the above helps, and thanks for your feedback!

Hi kat, check out the hintline (bottom of the user interface), it will list the modifiers/shortcuts you can use with whichever tool you have selected. In the case of the Paint Mixer Brush, you can use L to load the brush and C to clean. Hope that helps!

Hi @Anila, are you sure they're actually blurry, or it is more likely that you're just seeing it without any sharpening applied? Affinity Photo does minimal enhancement when developing RAW files. You can either add sharpening during the development stage (Details panel>Detail Refinement) or once you've developed your RAW image and moved to the Photo Persona (Layer>New Live Filter Layer>Unsharp Mask). I understand a concern might be that you're losing detail—don't worry as this isn't the case, the image would look soft in most RAW development applications if you removed the sharpening. However, if your images are actually really soft (e.g. sharpening doesn't solve the issue) it would be very useful if you could provide a sample RAW file to test with. A private Dropbox link can be provided if you don't want the file to be seen publicly. Hope that helps!

Hi @1drey and @Jeremy See, please check out this post I've made in another thread as I'd love to get some feedback about whether these macros can solve your issue: Alternatively, here's a download link: http://jamesritson.co.uk/downloads/macros/jr_360.zip And a copy/paste of the post: On macOS, you can drag-drop them straight into Affinity Photo and it will automatically import them and open the Library panel. On Windows, you'll need to open the Library panel manually (View>Studio>Library) then click the top right icon and Import Macros. There are four macros: Tone Map SDR (seam aware) Local Contrast (seam aware) Clarity (seam aware) Inpaint alpha (transparent) areas I've tested on a variety of imagery from HDRI Haven, HDR Labs, some customer files and my own 360 images. By and large, the three seamless macros will work very well. The only problematic one may be Clarity, in which case you'll end up with a seam that runs through 1/4 of the edge rather than all the way around, so it's much easier to retouch. I have found that Clarity in particular may also expose any existing stitching errors that usually wouldn't be obvious without heavy pixel modification, so bear that in mind as well. Hope the above helps!

Hi Scott, not an intentional plug but I wrote a procedural texture filter that acts as a green screen keyer—it's a lot quicker than creating manual selections and you can adjust the matte spill, antialiasing and fringing (green saturation). It's in my Workflow Bundle pack (https://forum.affinity.serif.com/index.php?/topic/100491-jr-workflow-bundle-shortcuts-macros-hdr-tools-brushes/) but I've attached it to this post instead. It'd be great to see if it works well for your imagery (I've only tested it on stock imagery with distinct green/blue backgrounds). Basically, you just run the macro on whichever layer (usually Background, the image layer) and then double click the Green Screen Key layer to access the controls. I remember I did a quick tweet with a video clip that shows it in action as well: Hope that helps! JR - Matting & Keying.afmacros

I agree with you, as controversial an opinion as that may be There appear to have been several threads about this lately—the main thing that needs to be said is that Affinity Photo's histogram does not represent luminance/luminosity. What you're seeing is the overlap/addition of the RGB channels. That's why it was changed from white, because users were mistaking it for luminosity. If you want a better representation of luminosity, check out the Intensity Waveform on the Scope panel (View>Studio>Scope). That gives you an IRE readout and an abstract representation of your image, which is much more useful for seeing where the tones in your image are. You can also utilise an RGB Parade for a more accurate idea of where your colour channels may be clipping.

Hi @AverytheFurry, looks like a colour management issue. The apps you're viewing the image in either won't be colour managed (looks like Photos isn't at all) or might be colour managed to a specific colour space. Being that it's for video streaming/recording, OBS might be assuming Rec.601 or Rec.709 colour primaries (or assuming sRGB and converting to Rec.601/Rec.709). As @GarryP mentioned, could you check what colour profile your document is using? To do this, make sure you have the View Tool selected (H on the keyboard) and look at the context toolbar readout. You'll have the pixel resolution followed by the colour profile. To ensure colours look as consistent as possible with apps that aren't colour managed, you're best off working in sRGB. You can either convert your document colour profile whilst working on it (Document>Convert Format / ICC Profile) or during export. To do it at export time, click the More dialog option and set ICC Profile to sRGB IEC 61966-2.2. If OBS is your intended destination, you might also want to try setting your document colour profile to Rec.709 (HD) or Rec.601 (SD) to see if that makes the result in OBS look consistent with what you're seeing in Affinity Photo. In the OBS Settings dialog, you've got the Advanced category where you can set the Colour Space. I believe it defaults to 601 but you can change this to 709. That said, from your screen grabs it looks like the result in OBS is very similar or the same as what you're seeing in Affinity Photo—which apps are you having issues with? [Edit] If you have a spare 10 minutes this colour management article on Spotlight may be of use (specifically the part near the end which covers document colour profiles: https://affinityspotlight.com/article/display-colour-management-in-the-affinity-apps/

Just to point something out—colour decontamination is actually applied to the previews (Overlay, Black/White and Transparent) and carries across when you output as a New layer or New layer with mask. You should always use one of these two options for cutouts/foreground extraction. Outputting to a selection or mask will forego colour decontamination because they don't alter the pixel content. Could I ask what your source file for this document was? I'm noticing some blocking and compression artefacting especially around the edges of the subject, suggesting the original image was quite noisy and was then compressed—not making excuses, it simply appears that Affinity Photo's selection refinement doesn't cope too well with compression artefacting. Hopefully the support team can take a more detailed look at this in the coming week!

I agree that it's not telegraphed very well, but essentially colour decontamination is linked to the issue @carl123 has mentioned as well as yourself in the above part I've quoted with using output to mask. Basically: Output to Selection/Output to Mask is for when you intend to mask an adjustment or filter layer, or perhaps create mask bounds for brush work. Because the pixel layer is not modified, no colour decontamination can be applied, so you only get the refined selection. Output to New layer/New layer with mask is for when you want to cut content out or isolate it from its background. Because the pixel content is modified, it means colour decontamination can be performed. The confusion may arise because the preview modes (Overlay, Transparent, Black/White) apply the colour decontamination procedure, whereas if you output to an active selection or mask there is of course no way to apply this because it requires modifying the pixel content. If you need to do this, you are better off outputting to a selection or mask as you mentioned. The artefacts are a result of discarding the background colour contribution and using only the foreground colours over the matted areas—the intention is for these to always be hidden by the mask (or discarded if you just output to a new layer). Should you choose to output as a new layer with mask, you also have other options to modify the mask: Add a Curves or Levels adjustment, drag it over the thumbnail of the mask (this will place it beneath the mask). Set the channel target to Alpha. You can now control the matte blending. Right click the mask layer and choose Refine Mask. Uncheck Matte edges (since the masking is already matted and you don't want to further matte the decontaminated areas) and set the preview mode to Transparent. You can then use Smooth, Feather and Ramp to further adjust the mask.

This isn't broken—the additional pixels you see have been treated for colour decontamination to optimise edge detail when the subject is cut out/isolated. As @walt.farrell mentioned above, you would not ordinarily see this because the mask is applied. If you were to take the destructive approach and only output as a New Layer you would never see these pixels, as they would simply be discarded.

Hi @Kutsche66, the difference you're seeing is expected regardless of whether you're in 16-bit or 32-bit, here's a quick explanation: When developing a RAW file, Photo knows the initial white balance (white point) and is able to change this which in turn offsets all the relative colour values. When entering the Develop Persona from a pixel layer or using the White Balance adjustment layer, the initial white balance is no longer known, plus the image has already been mapped to a colour space and relative white point. Therefore, rather than presenting a Kelvin scale, you instead have percentages—here, the adjustment is simply performing an offset based on an arbitrary white point, and so the result will look different (this includes the usage of the colour picker). That said, the scaling of the percentage version is not set in stone: it could be modified to more closely match the scale of the Kelvin version. It would however have an implication on existing documents that use the White Balance adjustment layer, so would require care if it was to be modified. There's an additional complexity with your image in that it requires a very dramatic white balance shift, e.g. if you use the white balance picker on the white coral the Kelvin value shifts to around 9000K. The adjustment version with its percentage slider can't go this far, so you won't be able to get the two to match. Hope that answers your question! I'll investigate the white balance scaling and see if it's possible to make some improvements here.

Hi Dave, you've got a blend range set on the Curves adjustment (linear 100% black to 0% white), this is probably what's causing it unless I'm missing something. Resetting the blend range then doing as you described (top left to clip the entire image) behaves as expected. Hope that helps!

Hey @SpartaPhoto, from my understanding you want to do the following: Blend the lighter doorway area from layer-1 into the composite of layer-2 Blend through only the luminance and not the colour information If this is correct, there are a couple of pointers: Setting the blend mode on the mask (e.g. Luminosity) will not do anything It looks like you have the layers the other way round in Affinity Photo when compared to your ON1 Photo Raw screen grabs (and presumably Photoshop as well?) To replicate the result you're getting in ON1 Photo Raw, just do the following: Put layer-1 above layer-2 Set layer-1's Blend Mode to Luminosity (the pixel layer, not the mask) Add a layer mask and invert it (Layer>Invert) Use the Paint Brush Tool to paint in over the doorway I've attached a quick video to demonstrate (see attached at bottom of post). Hopefully this is what you're trying to achieve? Note there is still a little bit of orange colour bleed in the top left of the doorway but this is consistent with the result from ON1 Photo Raw. You could always add a quick HSL adjustment layer, desaturate, invert (Layer>Invert again) and paint back in over that area if it concerns you. Hope that helps! luminosity_masking_trimmed.mp4

Hi @travisrogers, yes, managed it! (Albeit with some very minor differences when zoomed in and doing a quick A/B comparison) When opened in Affinity Photo, the reference TIFF is being colour managed from sRGB to the display profile, so for this challenge I'm assuming this result is the correct one. I'm also using ICC Display Transform and not OCIO Display Transform with the 32-bit EXR. You can append the colour space to the file name (e.g. "filename aces") and Photo will convert from that colour space to scene linear, but in practice with the file provided it made little to no difference. The next step is to add an OCIO adjustment layer and go from scene linear to out_srgb (the pointer/friendly name for this is "Output - sRGB"). Finally, to match the look of the 16-bit TIFF, we need to add a simple gamma transform—this is because ICC Display Transform uses the display profile's non-linear transform to ensure parity with the results you'll get when you export to a non-linear image format like JPEG. A Levels adjustment isn't flexible enough here, so the easiest way to achieve this is to add a live Procedural Texture filter and do a power transform of 2.2 for each colour channel. I've attached a screen grab below to illustrate: And that should be it! I've attached a side-by-side comparison. This is the best match I've been able to achieve so far..

Hi @johncena, there are a couple of things to break down in this reply: Firstly, regarding your OCIO configuration not working, which configuration are you using? If you've just copied the blender OCIO configuration from the blender directory, maybe try putting it somewhere that doesn't require elevated privileges (e.g. Documents folder or somewhere other than Program Files). Regarding the actual filmic transform, there's a little caveat you need to be aware of with blender's OCIO configuration. When the configuration is successfully loaded and you open an EXR file, the 32-bit Preview Panel will actually use "None" as the initial device transform, which gives you an unmanaged linear light view. If you intend to edit your document and export back to EXR, you'll want to switch this over to sRGB and choose Filmic as the view transform. This whole process is non-destructive and only applied to the view, though—the colour values in your document are not altered. If Affinity Photo is your final destination and you intend to export to a non-linear format (e.g. JPEG/TIFF/PNG), I'd recommend switching over to ICC Display Transform instead, which will apply a non-linear view transform based on the document's colour profile. This will then ensure parity with how the exported image will look. At this point, however, everything will look too bright and washed out compared to blender's view. You'll need to add an OCIO adjustment layer going from scene linear (usually just "Linear") to Filmic sRGB. Then you'll need to add a second OCIO adjustment layer going from sRGB to Linear. A bit strange, I know, but essentially your document is still in a linear colour space—there's just a non-linear view transform being applied when it's presented to screen. If you're unsure, I did a tutorial on this here: Hope the above helps!

Hi @Arifin, are you sure? Look at the Colour panel where you have the RGB sliders, they appear incorrect as well. Is there any chance you could screenshot the Colour tab under System Preferences>Displays and attach it here? (Shift+CMD+4 then tap Spacebar and click on the window to screenshot it). Affinity Photo will be colour managing based on whatever display profile you have active... are there any other details you could provide, e.g. what monitor type you're using?

Hi Dziga, the option you're looking for is under Preferences>Colour, near the bottom you have several OpenEXR options. Enable "Associate OpenEXR alpha channels" to premultiply the alpha channels into the .RGB pixel layers. Hope that helps! Also, there's no need to use the flood select tool. As @firstdefence mentioned above, you can also right click any layer including the greyscale alpha layer and choose Rasterise to Mask to convert it to a usable mask.

Hi @bvz, apologies as I've also been out of office. Looks like you have messaging turned off (or I'm blind and can't find the option to PM you ) so I've sent a file request to your email address, hope that's OK. Look forward to seeing the config and trying to puzzle this one out..

Hi, are you viewing on two different monitors? The screen grab on the left is significantly sharper as if it were on a high DPI monitor (look at the text), whereas the screen grab on the right looks much softer, as if it were captured on a standard monitor...

Hey @bvz, I've just had a quick test with OCIO on Windows and it appears to be working fine with multiple OCIO configurations (blender, Nuke, ACES etc). My first suggestion would have been to check that your document was in 32-bit but you've already covered that, so we should then move onto what your configuration actually contains. Affinity Photo exposes display/view transforms in the 32-bit preview panel—these do have to be explicitly defined as view transforms and not just general colour space transforms however. Even if you don't have any view transforms, you can still apply the colour space transforms by using an OCIO Adjustment layer (under Layer>New Adjustment Layer). Do your colour space transforms show up in this list? If so, you can likely choose Linear (or whatever role is defined as scene linear) as the Source, and whichever colour space you wish to transform to as the Destination. Hope that helps—if you do have view transforms defined in the configuration and it looks like Affinity Photo isn't picking them up, are you able to attach the OCIO configuration file and relevant directories? (If privacy is required we can send you an internal file sharing link)

Hi all, thanks for your feedback (it is being read!) in this thread. We're rounding off this new set of tutorials with these three: Cropping Straightening Images Procedural Texture: Tone Mapping HDR to SDR More tutorials will of course be on the way in the future—thanks again and hope you find them useful. As per usual, the first post has been updated too.

R16 is a single channel format that contains red channel information—it's typically used for height maps in landscape creation software/game engines etc. Affinity Photo doesn't have any direct R16 export capabilities. However, you could try the following @davide445 : Flatten your document if you have layer work (Document>Flatten). On the Channels panel, scroll down and right click Pixel Green then choose Clear. Do the same with Pixel Blue. Note that Pixel will be whatever your layer is named (so it might be Background if you haven't flattened your document). Ignore the Composite channels at the top as they won't offer the option to clear the channel data. You should now be left with just red channel information. Go to File>Export and choose the PNG format. Click the More button and find Pixel format. Choose Greyscale 16-bit from the dropdown and then export. Your mileage may vary with this—some people have reported success importing these 16-bit greyscale PNGs into the software they're using. Alternatively, if you don't specifically need the greyscale bitmap and just wanted to export an RGB image with only the red channel data, ignore the PNG export steps and just use whatever format you wish. Hope that helps!

Hey again, these new videos have been trickling out on YouTube over the past few days, so I've updated the first post with them: Pixel vs Image layers Using Matte ID render passes for masking Lock children (Masking) As always, hope you find them useful!