Ammonia emissions produce negative environmental and human health impacts with largest emissions originating from agriculture. Especially in countries with high livestock density, the majority originate from animal housing and application to fields. Measuring total emissions from multiple heterogenous source structures such as farms and waste treatment facilities can be challenging due to losses from transport, deposition, and chemical transformation. Previous studies have shown that quantifying net fluxes at this scale can be achieved by combining concentration measurements up- and downwind of the structures with inverse dispersion modelling to calculate the emissions from a defined source area. However, this method underestimates total emissions, as it does not account for deposition loss, which must be modelled and can introduce large uncertainties (

This study examines methane (CH4) emission factors from biogas and wastewater treatment plants, based on primary and secondary data collected from 109 facilities. Primary emission data were measured at 19 facilities representing prevalent plant configurations across Europe. Statistical analysis highlights two categorical variables, namely primary feedstock and plant size, expressed as CH4 production (<= 250 kg h(-1): small and medium-sized plants, > 250 kg h(-1): large plants), each of which has a significant impact on whole-site CH4 emissions. Additionally, digestate storage (gastight vs. not-gastight) has a meaningful effect when considering CH4 production as a continuous variable in the statistical analysis. Our results indicate that wastewater treatment plants have the highest average CH4 losses (7.0 % of CH4 produced, n = 31 or 0.10 kg population equivalent (PE)(-1) yr(-1), n = 28), followed by manure-based plants (3.7 %, n = 49), biowaste treatment facilities (2.8 %, n = 11) and energy crop-processing plants (1.9 %, n = 14). Furthermore, small and medium-sized plants have elevated emissions (5.6 %, n = 67) compared to larger counterparts (2.2 %, n = 42), primarily attributed to the absence of gastight digestate storage. Emissions tend to be lower with gastight digestate storage (2.7 %, n = 61) than not-gastight storage options (6.2 %, n = 48). Emission factors were determined for normal operating conditions, with a further investigation into other-than-normal operating conditions revealing temporal or constant emission peaks in eight out of 19 facilities. These peaks, suggesting potential areas for targeted mitigation strategies, were attributed to pressure relief valves, flare ignition problems and major leakages.

Volumetric modeling and neural radiance field representations have revolutionized 3D face capture and photorealistic novel view synthesis. However, these methods often require hundreds of multi-view input images and are thus inapplicable to cases with less than a handful of inputs. We present a novel volumetric prior on human faces that allows for high-fidelity expressive face modeling from as few as three input views captured in the wild. Our key insight is that an implicit prior trained on synthetic data alone can generalize to extremely challenging real-world identities and expressions and render novel views with fine idiosyncratic details like wrinkles and eyelashes. We leverage a 3D Morphable Face Model to synthesize a large training set, rendering each identity with different expressions, hair, clothing, and other assets. We then train a conditional Neural Radiance Field prior on this synthetic dataset and, at inference time, fine-tune the model on a very sparse set of real images of a single subject. On average, the fine-tuning requires only three inputs to cross the synthetic-to-real domain gap. The resulting personalized 3D model reconstructs strong idiosyncratic facial expressions and outperforms the state-of-the-art in high-quality novel view synthesis of faces from sparse inputs in terms of perceptual and photo-metric quality.

We introduce a novel framework for 3D human avatar generation and personalization, leveraging text prompts to enhance user engagement and customization. Central to our approach are key innovations aimed at overcoming the challenges in photo-realistic avatar synthesis. Firstly, we utilize a conditional Neural Radiance Fields (NeRF) model, trained on a large-scale unannotated multi-view dataset, to create a versatile initial solution space that accelerates and diversifies avatar generation. Secondly, we develop a geometric prior, leveraging the capabilities of Text-to-Image Diffusion Models, to ensure superior view invariance and enable direct optimization of avatar geometry. These foundational ideas are complemented by our optimization pipeline built on Variational Score Distillation (VSD), which mitigates texture loss and over-saturation issues. As supported by our extensive experiments, these strategies collectively enable the creation of custom avatars with unparalleled visual quality and better adherence to input text prompts. You can find more results and videos in our website: syntec-research.github.io/MagicMirror .

Emissions from agricultural sources substantially contribute to global warming. The inverse dispersion method (IDM) has been successfully used for emission measurements from various agricultural sources. The IDM has also been validated in multiple studies with artificial gas releases mostly in open fields. Release experiments from buildings have rarely been conducted and were partly affected by additional nearby sources of the target gas. Specific release studies for naturally ventilated animal housings are lacking. In this study, a known and predefined amount of methane (CH4) was released from an artificial source inside a barn that mimicked a naturally ventilated dairy housing, and IDM recovery rates, using a backward Lagrangian stochastic (bLS) model, were determined. For concentration measurements, open-path devices (OPs) with a path length of 110 m were placed in a downwind direction of the barn at fetches of 2.0h, 5.3h, 8.6h, and 12h (h equals the height of the highest obstacle), and a 3D ultrasonic anemometer (UA) was placed in the middle of the first three OP paths. Upwind of the barn, an additional OP and a UA were installed. The median IDM recovery rates determined with the UA placed upwind of the barn and the downwind OP ranged between 0.55–0.75. It is concluded that, for the present study case, the effect of the building and a tree in the main wind axis led to a systematic underestimation of the IDM-derived emission rate probably due to deviations in the wind field and turbulent dispersion from the underlying assumptions of the used dispersion model.

Ammonia (NH3) emissions negatively impact air, soil, and water quality, hence human health and biodiversity. Significant emissions, including the largest sources, originate from single or multiple structures, such as livestock facilities and wastewater treatment plants (WWTPs). The inverse dispersion method (IDM) is effective in measuring total emissions from such sources, although depositional loss between the source and point of measurement is often not accounted for. We applied IDM with a deposition correction to determine total emissions from a representative dairy housing and WWTP during several months in autumn and winter in Switzerland. Total emissions were 1.19 ± 0.48 and 2.27 ± 1.53 kg NH3 d-1 for the dairy housing and WWTP, respectively, which compared well with literature values, despite the paucity of WWTPs data. A concurrent comparison with an inhouse tracer ratio method at the dairy housing indicated an offset of the IDM emissions by <20%. Diurnal emission patterns were evident at both sites mostly driven by changes in air temperature with potential lag effects such as following sludge agitation. Modelled deposition corrections to adjust the concentration loss detected at the measurement point with the associated footprint were 22-28% of the total emissions and the cumulative fraction of deposition to emission modelled with distance from the source was between 7-12% for the measurement distances (60-150 m). Although estimates of depositional loss were plausible, the approach is still connected with substantial uncertainty, which calls for future validation measurements. Longer measurement periods encompassing more management activities and environmental conditions are required to assess predictor variable importance on emission dynamics. Combined, IDM with deposition correction will allow the determination of emission factors at reduced efforts and costs, thereby supporting the development and assessment of emission reducing methods and expand the data availability for emission inventories.

These data contain ammonia concentration measurements using line-integrated miniDOAS instruments and emission calculations using an inverse dispersion method applying the bLS model at a wastewater treatment plant during several weeks in September and October 2019 in Switzerland. The data are described in a manuscript titled "Ammonia emissions from a dairy housing and wastewater treatment plant quantified with an inverse dispersion method accounting for deposition loss" submitted to the Journal of Air & Waste Management in 2023. These measurements were funded by the Swiss Federal Office for the Environment (Contract 06.0091.PZ / R281-0748).

These data contain ammonia concentration measurements using line-integrated miniDOAS instruments and emission calculations using an inverse dispersion method applying the bLS model at an experimental dairy housing operated by Agroscope in Tänikon, Switzerland during several months from September to December 2018. The data are described in a manuscript titled "Ammonia emissions from a dairy housing and wastewater treatment plant quantified with an inverse dispersion method accounting for deposition loss" submitted to the Journal of Air & Waste Management in 2023. These measurements were funded by the Swiss Federal Office for the Environment (Contract 00.5082.P2I R254-0652).

High-fidelity, photorealistic 3D capture of a human face is a long-standing problem in computer graphics – the complex material of skin, intricate geometry of hair, and fine scale textural details make it challenging. Traditional techniques rely on very large and expensive capture rigs to reconstruct explicit mesh geometry and appearance maps, and are limited by the accuracy of hand-crafted reflectance models. More recent volumetric methods (e.g., NeRFs) have enabled view-synthesis and sometimes relighting by learning an implicit representation of the density and reflectance basis, but suffer from artifacts and blurriness due to the inherent ambiguities in volumetric modeling. These problems are further exacerbated when capturing with few cameras and light sources. We present a novel technique for high-quality capture of a human face for 3D view synthesis and relighting using a sparse, compact capture rig consisting of 15 cameras and 15 lights. Our method combines a neural volumetric representation with traditional mesh reconstruction from multiview stereo. The proxy geometry allows us to anchor the 3D density field to prevent artifacts and guide the disentanglement of intrinsic radiance components of the face appearance such as diffuse and specular reflectance, and incident radiance (shadowing) fields. Our hybrid representation significantly improves the state-of-the-art quality for arbitrarily dense renders of a face from desired camera viewpoint as well as environmental, directional, and near-field lighting.

A unique challenge in creating high-quality animatable and relightable 3D avatars of real people is modeling human eyes, particularly in conjunction with the surrounding periocular face region. The challenge of synthesizing eyes is multifold as it requires 1) appropriate representations for the various components of the eye and the periocular region for coherent viewpoint synthesis, capable of representing diffuse, refractive and highly reflective surfaces, 2) disentangling skin and eye appearance from environmental illumination such that it may be rendered under novel lighting conditions, and 3) capturing eyeball motion and the deformation of the surrounding skin to enable re-gazing. These challenges have traditionally necessitated the use of expensive and cumbersome capture setups to obtain high-quality results, and even then, modeling of the full eye region holistically has remained elusive. We present a novel geometry and appearance representation that enables high-fidelity capture and photorealistic animation, view synthesis and relighting of the eye region using only a sparse set of lights and cameras. Our hybrid representation combines an explicit parametric surface model for the eyeball surface with implicit deformable volumetric representations for the periocular region and the interior of the eye. This novel hybrid model has been designed specifically to address the various parts of that exceptionally challenging facial area - the explicit eyeball surface allows modeling refraction and high frequency specular reflection at the cornea, whereas the implicit representation is well suited to model lower frequency skin reflection via spherical harmonics and can represent non-surface structures such as hair (i.e. eyebrows) or highly diffuse volumetric bodies (i.e. sclera), both of which are a challenge for explicit surface models. Tightly integrating the two representations in a joint framework allows controlled photoreal image synthesis and joint optimization of both the geometry parameters of the eyeball and the implicit neural network in continuous 3D space. We show that for high-resolution close-ups of the human eye, our model can synthesize high-fidelity animated gaze from novel views under unseen illumination conditions, allowing to generate visually rich eye imagery.

Wastewater treatment plants (WWTPs) and biogas plants (BGPs) are significant sources of methane (CH4), with a combined share of around 40% within the waste sector of the Swiss national emission inventory. We conducted whole-plant CH4 emission measurements at two WWTPs and four agricultural BGPs in Switzerland using the inverse dispersion method (IDM). This involved open-path concentration measurements up- and downwind of the plant in combination with a backward Lagrangian stochastic (bLS) model. WWTPs in particular consist of multiple CH4 sources with different areas and emission strengths. For the combination of the individual emission sources in the bLS modelling, three different calculation approaches with different levels of detail were applied: (i) single source over enveloping polygon area, (ii) uniform emission density for all individual source areas, (iii) specified relative weighting of individual sources based on literature data. Average CH4 emissions for WWTP-1 and WWTP-2 were 0.82 kg h−1 and 0.61 kg h−1 and scaled to population equivalents (PE) 166 g PE−1 y−1 and 381 g PE−1 y−1, respectively. BGPs CH4 emissions varied between 0.39 kg h−1 and 2.22 kg h−1, corresponding to less than 5% of the plants’ CH4 production. The highest numbers were due to measurements during other than normal operating conditions. The emissions of WWTPs and BGPs comply with literature values. Approach (iii) with source weighting led to a difference of up to 43% for the two WWTPs compared to the assumption of uniform emissions. Furthermore, we demonstrate how multiple open-path concentration measurements can be combined and how the measurements can be corrected for nearby external CH4 sources not belonging to the investigated plants. The results of the present study contribute to improved emission data from the waste sector.

Traditional 3D morphable face models (3DMMs) provide fine-grained control over expression but cannot easily capture geometric and appearance details. Neural volumetric representations approach photorealism but are hard to animate and do not generalize well to unseen expressions. To tackle this problem, we propose IMavatar (Implicit Morphable avatar), a novel method for learning implicit head avatars from monocular videos. Inspired by the fine-grained control mechanisms afforded by conventional 3DMMs, we represent the expression- and pose-related deformations via learned blendshapes and skinning fields. These attributes are pose-independent and can be used to morph the canonical geometry and texture fields given novel expression and pose parameters. We employ ray marching and iterative root-finding to locate the canonical surface intersection for each pixel. A key contribution is our novel analytical gradient formulation that enables end-to-end training of IMavatars from videos. We show quantitatively and qualitatively that our method improves geometry and covers a more complete expression space compared to state-of-the-art methods. Code and data can be found at https://ait.ethz.ch/projects/2022/IMavatar/.

Open-path measurements of methane (CH4) with the use of GasFinder systems (Boreal Laser Inc, Edmonton Canada) have been frequently used for emission estimation with the inverse dispersion method (IDM), particularly from agricultural sources. It is common to many IDM applications that the concentration enhancement related to CH4 sources is small, typically between 0.05 and 0.5 ppm, and accurate measurements of CH4 concentrations are needed at concentrations close to ambient levels. The GasFinder3-OP (GF3) device for open-path CH4 measurements is the latest version of the commercial GasFinder systems by Boreal Laser Inc. We investigated the uncertainty of six GF3 devices from side-by-side intercomparison measurements and comparisons to a closed-path quantum cascade laser device. The comparisons were made at near-ambient levels of CH4 (85 % of measurements below 2.5 ppm) with occasional phases of elevated concentrations (max. 8.3 ppm). Relative biases as high as 8.3 % were found, and a precision for half-hourly data between 2.1 and 10.6 ppm-m (half width of the 95 % confidence interval) was estimated. These results deviate from the respective manufacturer specifications of 2 % and 0.5 ppm-m. Intercalibration of the GF3 devices by linear regression to remove measurement bias was shown to be of limited value due to drifts and step changes in the recorded GF3 concentrations.

Methane (CH4) emissions from dairy housings, mainly originating from enteric fermentation of ruminating animals, are a significant source of greenhouse gases. The quantification of emissions from naturally ventilated dairy housings is challenging due to the spatial distribution of sources (animals, housing areas) and variable air exchange. The inverse dispersion method (IDM) is a promising option, which is increasingly used to determine gaseous emissions from stationary sources, as it offers high flexibility in the application at reasonable costs. We used a backward Lagrangian stochastic model combined with concentration measurements by open-path tunable diode laser spectrometers placed up- and downwind of a naturally ventilated housing with 40 dairy cows to determine the CH4 emissions. The average emissions per livestock unit (LU) were 317 (+/- 44) g LU-1 d(-1) and 267 (+/- 43) g LU-1 d(-1) for the first and second campaign, in September - October and November - December, respectively. For each campaign, inhouse tracer ratio measurements (iTRM) were conducted in parallel during two subperiods. For simultaneous measurements, IDM showed average emissions which were lower by 8% and 1% than that of iTRM, respectively, for the two campaigns. The differences are within the uncertainty range of any of the two methods. The IDM CH4 emissions were further analysed by wind direction and atmospheric stability and no differences in emissions were found. Overall, IDM showed its aptitude to accurately determine CH4 emissions from dairy housings or other stationary sources if the site allows adequate placement of sensors up- and downwind in the prevailing wind direction. To acquire reliable emission data, depending on the data loss during measurements due to quality filtering or instrument failure, a measuring time of at least 10 days is required.

Super-resolution (SR) is by definition ill-posed. There are infinitely many plausible high-resolution variants for a given low-resolution natural image. Most of the current literature aims at a single deterministic solution of either high reconstruction fidelity or photo-realistic perceptual quality. In this work, we propose an explorative facial super-resolution framework, DeepSEE, for Deep disentangled Semantic Explorative Extreme super-resolution. To the best of our knowledge, DeepSEE is the first method to leverage semantic maps for explorative super-resolution. In particular, it provides control of the semantic regions, their disentangled appearance and it allows a broad range of image manipulations. We validate DeepSEE on faces, for up to 32x magnification and exploration of the space of super-resolution. Our code and models are available at: https://mcbuehler.github.io/DeepSEE/

Deep generative models can synthesize photorealistic images of human faces with novel identities. However, a key challenge to the wide applicability of such techniques is to provide independent control over semantically meaningful parameters: appearance, head pose, face shape, and facial expressions. In this paper, we propose VariTex - to the best of our knowledge the first method that learns a variational latent feature space of neural face textures, which allows sampling of novel identities. We combine this generative model with a parametric face model and gain explicit control over head pose and facial expressions. To generate complete images of human heads, we propose an additive decoder that adds plausible details such as hair. A novel training scheme enforces a pose-independent latent space and in consequence, allows learning a one-to-many mapping between latent codes and pose-conditioned exterior regions. The resulting method can generate geometrically consistent images of novel identities under fine-grained control over head pose, face shape, and facial expressions. This facilitates a broad range of downstream tasks, like sampling novel identities, changing the head pose, expression transfer, and more.

Inverse dispersion modelling (IDM) using a backward Lagrangian stochastic (bLS) dispersion model has been successfully applied to quantify emissions from confined ground sources e.g. as for ammonia (NH3) loss after manure spreading. The most widely used bLS model for emission measurements of NH3 and methane (CH4) from agricultural sources such as lagoons and livestock buildings is based on Flesch et al. (2004). For such applications, the model assumptions of a diffusive ground source within a homogeneous turbulence field, which implies absence of obstacles as e.g. buildings disturbing the flow, is clearly not fulfilled. It remains unclear to what extend these violations introduce bias into the emission estimates. Further, the model by Flesch et al. does not include deposition removal, which for NH3, can induce an underestimation of the emission from the source (Häni et al., 2018). Häni et al. extended the standard bLS calculation model with an optional dry deposition mechanism. In a field campaign between mid-September and mid-December 2018, CH4 and NH3 emissions from a natural ventilated dairy housing with 40 cows were quantified using the IDM method with the bLS model by Häni et al. (2018). From the three-month period, results for 63 measurement days at 30-minute resolution were evaluated and thereof 71% of the data points were discarded from the emission calculation due to inapplicable turbulence conditions or instrument failure. NH3 and CH4 concentrations were analysed with open-path instruments (NH3: miniDOAS,; CH4: GasFinder, Boreal Laser, Inc., Edmonton, Alberta, Canada) (aligned in parallel) with 50 m path lengths (distance between sensor and reflector). During part of the field campaign (24 days), simultaneous in-house measurements of CH4 and NH3 emissions using the tracer ratio method (iTM) (SF6 and SF5CF3, Mohn et al., 2018) were conducted and results compared with the estimates retrieved by the IDM method. Overall, the results from the IDM method compare well to the results of the in-house measurements, with mean daily emissions of 18.3 kg CH4/d (IDM) and 17.9 kg CH4/d (iTM) and 1.08 kg NH3/d (IDM) and 1.56 kg NH3/d (iTM), respectively. Regarding NH3, the IDM method was run without the inclusion of a dry deposition mechanism. First results from IDM calculations with the inclusion of dry deposition indicate, that dry deposition modelling may explain the difference in NH3 emissions between the IDM method and the iTM. References Flesch, T. K., Wilson, J. D., Harper, L. A., Crenna, B. P., and Sharpe, R. R.: Deducing ground-to-air emissions from observed trace gas concentrations: A field trial, J. Appl. Meteorol., 43, 487–502, 2004. Häni, C., Flechard, C., Neftel, A., Sintermann, J., and Kupper, T.: Accounting for Field-Scale Dry Deposition in Backward Lagrangian Stochastic Dispersion Modelling of NH3 Emissions, Atmosphere, 9, 146, 2018. Mohn, J., Zeyer, K., Keck, M., Keller, M., Zähner, M., Poteko, J., Emmenegger, L., and Schrade, S.: A dual tracer ratio method for comparative emission measurements in an experimental dairy housing, Atmospheric Environment, 179, 12–22, 2018.

Quantification of gaseous emissions from waste water treatment plants (WWTPs) is challenging due to the heterogeneity of the emissions in space and time. The inverse dispersion method (IDM) using concentration and turbulence measurements in combination with a backward Lagrangian stochastic (bLS) dispersion model based on Flesch et al. (2004) is a promising option. It is increasingly used to determine gaseous emissions from confined sources (Flesch et al., 2009; VanderZaag et al., 2014), as it offers high flexibility at reasonable costs. For the application on WWTPs the bLS model assumption of spatially homogeneous turbulence, which implies absence of obstacles as buildings and trees that disturbe the flow, is often not fulfilled. However, studies showed that with the correct instrument setup and data filtering the bLS can be used for emission estimates. Methane emissions from two WWTPs of different type and size were quantified using the IDM with the bLS model. Methane concentrations were analysed with open-path tunable diode laser spectrometers (GasFinder, Boreal Laser, Inc., Edmonton, Alberta, Canada) placed up- and downwind of the source. At each site at least 20 days of measurements averaged to 30-minutes intervals are available. Here we present first results from these two WWTPs emission estimates. References Flesch, T. K., Wilson, J. D., Harper, L. A., Crenna, B. P., and Sharpe, R. R.: Deducing ground-to-air emissions from observed trace gas concentrations: A field trial, J. Appl. Meteorol., 43, 487–502, doi:10.1175/1520-0450(2004)043<0487:DGEFOT>2.0.CO;2, 2004. Flesch, T. K., Harper, L. A., Powell, J. M., and Wilson, J. D.: Inverse-dispersion calculation of ammonia emissions from Wisconsin dairy farms, Trans. ASABE, 52, 253–265, doi:10.13031/2013.25946, 2009. VanderZaag, A. C., Flesch, T. K., Desjardins, R. L., Baldé, H., and Wright, T.: Measuring methane emissions from two dairy farms: Seasonal and manure-management effects, Agricultural and Forest Meteorology, 194, 259–267, doi:10.1016/j.agrformet.2014.02.003, 2014.

Storage of slurry is an important emission source for ammonia (NH3), nitrous oxide (N2O), methane (CH4), carbon dioxide (CO2) and hydrogen sulfide (H2S) from livestock production. Therefore, this study collected published emission data from stored cattle and pig slurry to determine baseline emission values and emission changes due to slurry treatment and coverage of stores. Emission data were collected from 120 papers yielding 711 records of measurements conducted at farm-, pilot- and laboratory-scale. The emission data reported in a multitude of units were standardized and compiled in a database. Descriptive statistics of the data from untreated slurry stored uncovered revealed a large variability in emissions for all gases. To determine baseline emissions, average values based on a weighting of the emission data according to the season and the duration of the emission measurements were constructed using the data from farm-scale and pilot-scale studies. Baseline emissions for cattle and pig slurry stored uncovered were calculated. When possible, it was further distinguished between storage in tanks without slurry treatment and storage in lagoons which implies solid-liquid separation and biological treatment. The baseline emissions on an area or volume basis are: for NH3:0.12 g M-2 h(-1) and 0.15 g M-2 h(-1) for cattle and pig slurry stored in lagoons, and 0.08 g M-2 h(-1) and 0.24 g M-2 h(-1) for cattle and pig slurry stored in tanks; for N2O: 0.0003 g M-2 h(-1) for cattle slurry stored in lagoons, and 0.002 g M-2 h(-1) for both slurry types stored in tanks; for CH4: 0.95 g M-2 h(-1) and 3.5 g M-2 h(-1) for cattle and pig slurry stored in lagoons, and 0.58 g M-2 h(-1) and 0.68 g M-2 h(-1) for cattle and pig slurry stored in tanks; for CO2: 6.6 g M-2 h(-1) and 0.3 g M-2 h(-1) for cattle and pig slurry stored in lagoons, and 8.0 g M-2 h(-1) for both slurry types stored in tanks; for H2S: 0.04 g M-2 h(-1) and 0.01 g M-2 h(-1) for cattle and pig slurry stored in lagoons. Related to total ammoniacal nitrogen (TAN), baseline emissions for tanks are 16% and 15% of TAN for cattle and pig slurry, respectively. Emissions of N2O and CH4 relative to nitrogen (N) and volatile solids (VS) are 0.13% of N and 0.10% of N and 2.9% of VS and 4.7% of VS for cattle and pig slurry, respectively. Total greenhouse gas emissions from slurry stores are dominated by CH4. The records on slurry treatment using acidification show a reduction of NH3 and CH4 emissions during storage while an increase occurs for N2O and a minor change for CO2 as compared to untreated slurry. Solid-liquid separation causes higher losses for NH3 and a reduction in CH4, N2O and CO2 emissions. Anaerobically digested slurry shows higher emissions during storage for NH3 while losses tend to be lower for CH4 and little changes occur for N2O and CO2 compared to untreated slurry. All cover types are found to be efficient for emission mitigation of NH3 from stores. The N2O emissions increase in many cases due to coverage. Lower CH4 emissions occur for impermeable covers as compared to uncovered slurry storage while for permeable covers the effect is unclear or emissions tend to increase. Limited and inconsistent data regarding emission changes with covering stores are available for CO2 and H2S. The compiled data provide a basis for improving emission inventories and highlight the need for further research to reduce uncertainty and fill data gaps regarding emissions from slurry storage.

Accurately labeled real-world training data can be scarce, and hence recent works adapt, modify or generate images to boost target datasets. However, retaining relevant details from input data in the generated images is challenging and failure could be critical to the performance on the final task. In this work, we synthesize person-specific eye images that satisfy a given semantic segmentation mask (content), while following the style of a specified person from only a few reference images. We introduce two approaches, (a) one used to win the OpenEDS Synthetic Eye Generation Challenge at ICCV 2019, and (b) a principled approach to solving the problem involving simultaneous injection of style and content information at multiple scales. Our implementation is available at https://github.com/mcbuehler/Seg2Eye.